2006-06-13 13:17:45 +00:00
|
|
|
/*
|
|
|
|
* Clutter.
|
|
|
|
*
|
|
|
|
* An OpenGL based 'interactive canvas' library.
|
|
|
|
*
|
|
|
|
* Authored By Matthew Allum <mallum@openedhand.com>
|
|
|
|
*
|
|
|
|
* Copyright (C) 2006 OpenedHand
|
|
|
|
*
|
|
|
|
* This library is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This library is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
|
|
* License along with this library; if not, write to the
|
|
|
|
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
|
|
|
|
* Boston, MA 02111-1307, USA.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
2006-06-21 22:34:25 +00:00
|
|
|
* SECTION:clutter-actor
|
2007-10-12 08:17:00 +00:00
|
|
|
* @short_description: Base abstract class for all visual stage actors.
|
|
|
|
*
|
2007-06-16 20:58:00 +00:00
|
|
|
* #ClutterActor is a base abstract class for all visual elements on the
|
|
|
|
* stage. Every object that must appear on the main #ClutterStage must also
|
|
|
|
* be a #ClutterActor, either by using one of the classes provided by
|
|
|
|
* Clutter, or by implementing a new #ClutterActor subclass.
|
2007-10-09 23:45:49 +00:00
|
|
|
*
|
2008-06-17 13:49:40 +00:00
|
|
|
* Every actor is a 2D surface positioned and optionally transformed
|
|
|
|
* in 3D space. The actor is positioned relative to top left corner of
|
|
|
|
* it parent with the childs origin being its anchor point (also top
|
|
|
|
* left by default).
|
|
|
|
*
|
|
|
|
* The actors 2D surface is contained inside its bounding box,
|
|
|
|
* described by the #ClutterActorBox structure:
|
2008-02-08 16:13:15 +00:00
|
|
|
*
|
|
|
|
* <figure id="actor-box">
|
|
|
|
* <title>Bounding box of an Actor</title>
|
|
|
|
* <graphic fileref="actor-box.png" format="PNG"/>
|
|
|
|
* </figure>
|
|
|
|
*
|
2008-06-17 13:49:40 +00:00
|
|
|
* The actor box represents the untransformed area occupied by an
|
|
|
|
* actor. Each visible actor that has been put on a #ClutterStage also
|
|
|
|
* has a transformed area, depending on the actual transformations
|
|
|
|
* applied to it by the developer (scale, rotation). Tranforms will
|
|
|
|
* also be applied to any child actors. Also applied to all actors by
|
|
|
|
* the #ClutterStage is a perspective transformation. API is provided
|
|
|
|
* for both tranformed and untransformed actor geometry information.
|
2008-06-12 16:34:50 +00:00
|
|
|
*
|
2008-06-17 13:49:40 +00:00
|
|
|
* The 'modelview' transform matrix for the actor is constructed from
|
2008-02-15 14:39:25 +00:00
|
|
|
* the actor settings by the following order of operations:
|
2007-11-28 12:23:31 +00:00
|
|
|
* <orderedlist>
|
|
|
|
* <listitem><para>Translation by actor x, y coords,</para></listitem>
|
2009-01-23 17:25:43 +00:00
|
|
|
* <listitem><para>Translation by actor depth (z),</para></listitem>
|
2007-11-28 12:23:31 +00:00
|
|
|
* <listitem><para>Scaling by scale_x, scale_y,</para></listitem>
|
|
|
|
* <listitem><para>Rotation around z axis,</para></listitem>
|
|
|
|
* <listitem><para>Rotation around y axis,</para></listitem>
|
|
|
|
* <listitem><para>Rotation around x axis,</para></listitem>
|
2009-01-23 17:25:43 +00:00
|
|
|
* <listitem><para>Negative translation by anchor point x,
|
|
|
|
* y,</para></listitem>
|
2008-06-17 13:49:40 +00:00
|
|
|
* <listitem><para>Rectangular Clip is applied (this is not an operation on
|
2008-06-12 16:34:50 +00:00
|
|
|
* the matrix as such, but it is done as part of the transform set
|
|
|
|
* up).</para></listitem>
|
2007-11-28 12:23:31 +00:00
|
|
|
* </orderedlist>
|
2007-10-09 23:45:49 +00:00
|
|
|
*
|
2008-06-12 16:34:50 +00:00
|
|
|
* An actor can either be explicitly sized and positioned, using the
|
|
|
|
* various size and position accessors, like clutter_actor_set_x() or
|
|
|
|
* clutter_actor_set_width(); or it can have a preferred width and
|
|
|
|
* height, which then allows a layout manager to implicitly size and
|
2008-06-17 13:49:40 +00:00
|
|
|
* position it by "allocating" an area for an actor. This allows for
|
2009-10-07 15:41:25 +00:00
|
|
|
* actors to be manipulated in both a fixed (or static) parent container
|
|
|
|
* (i.e. children of #ClutterGroup) and a more automatic (or dynamic)
|
2008-06-17 13:49:40 +00:00
|
|
|
* layout based parent container.
|
2008-06-12 16:34:50 +00:00
|
|
|
*
|
|
|
|
* When accessing the position and size of an actor, the simple accessors
|
|
|
|
* like clutter_actor_get_width() and clutter_actor_get_x() will return
|
|
|
|
* a value depending on whether the actor has been explicitly sized and
|
|
|
|
* positioned by the developer or implicitly by the layout manager.
|
|
|
|
*
|
|
|
|
* Depending on whether you are querying an actor or implementing a
|
|
|
|
* layout manager, you should either use the simple accessors or use the
|
|
|
|
* size negotiation API.
|
|
|
|
*
|
2008-06-17 13:49:40 +00:00
|
|
|
* Clutter actors are also able to receive input events and react to
|
|
|
|
* them. Events are handled in the following ways:
|
|
|
|
*
|
2007-10-10 14:51:25 +00:00
|
|
|
* <orderedlist>
|
|
|
|
* <listitem><para>Actors emit pointer events if set reactive, see
|
|
|
|
* clutter_actor_set_reactive()</para></listitem>
|
2007-11-18 23:24:55 +00:00
|
|
|
* <listitem><para>The stage is always reactive</para></listitem>
|
2007-10-15 16:50:59 +00:00
|
|
|
* <listitem><para>Events are handled by connecting signal handlers to
|
|
|
|
* the numerous event signal types.</para></listitem>
|
2007-10-10 14:51:25 +00:00
|
|
|
* <listitem><para>Event handlers must return %TRUE if they handled
|
2007-11-23 17:12:27 +00:00
|
|
|
* the event and wish to block the event emission chain, or %FALSE
|
2007-10-10 14:51:25 +00:00
|
|
|
* if the emission chain must continue</para></listitem>
|
|
|
|
* <listitem><para>Keyboard events are emitted if actor has focus, see
|
2007-11-23 17:12:27 +00:00
|
|
|
* clutter_stage_set_key_focus()</para></listitem>
|
|
|
|
* <listitem><para>Motion events (motion, enter, leave) are not emitted
|
|
|
|
* if clutter_set_motion_events_enabled() is called with %FALSE.
|
|
|
|
* See clutter_set_motion_events_enabled() documentation for more
|
|
|
|
* information.</para></listitem>
|
|
|
|
* <listitem><para>Once emitted, an event emission chain has two
|
2008-02-15 14:39:25 +00:00
|
|
|
* phases: capture and bubble. An emitted event starts in the capture
|
|
|
|
* phase (see ClutterActor::captured-event) beginning at the stage and
|
|
|
|
* traversing every child actor until the event source actor is reached.
|
|
|
|
* The emission then enters the bubble phase, traversing back up the
|
|
|
|
* chain via parents until it reaches the stage. Any event handler can
|
|
|
|
* abort this chain by returning %TRUE (meaning "event handled").
|
|
|
|
* </para></listitem>
|
2007-12-17 15:47:19 +00:00
|
|
|
* <listitem><para>Pointer events will 'pass through' non reactive
|
|
|
|
* overlapping actors.</para></listitem>
|
2007-10-25 14:36:37 +00:00
|
|
|
* </orderedlist>
|
2008-02-08 13:52:11 +00:00
|
|
|
*
|
|
|
|
* <figure id="event-flow">
|
|
|
|
* <title>Event flow in Clutter</title>
|
|
|
|
* <graphic fileref="event-flow.png" format="PNG"/>
|
|
|
|
* </figure>
|
|
|
|
*
|
2008-02-08 16:13:15 +00:00
|
|
|
* Every '?' box in the diagram above is an entry point for application
|
|
|
|
* code.
|
2008-02-15 14:39:25 +00:00
|
|
|
*
|
2008-06-17 13:49:40 +00:00
|
|
|
* For implementing a new custom actor class, please read <link
|
2008-02-15 14:39:25 +00:00
|
|
|
* linkend="clutter-subclassing-ClutterActor">the corresponding section</link>
|
|
|
|
* of the API reference.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
|
2007-11-23 13:11:10 +00:00
|
|
|
/**
|
|
|
|
* CLUTTER_ACTOR_IS_MAPPED:
|
2009-05-05 14:37:41 +00:00
|
|
|
* @a: a #ClutterActor
|
2007-11-23 13:11:10 +00:00
|
|
|
*
|
|
|
|
* Evaluates to %TRUE if the %CLUTTER_ACTOR_MAPPED flag is set.
|
|
|
|
*
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* Means "the actor will be painted if the stage is mapped."
|
|
|
|
*
|
|
|
|
* %TRUE if the actor is visible; and all parents with possible exception
|
|
|
|
* of the stage are visible; and an ancestor of the actor is a toplevel.
|
|
|
|
*
|
|
|
|
* Clutter auto-maintains the mapped flag whenever actors are
|
|
|
|
* reparented or shown/hidden.
|
|
|
|
*
|
2007-11-23 13:11:10 +00:00
|
|
|
* Since: 0.2
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
|
|
|
* CLUTTER_ACTOR_IS_REALIZED:
|
2009-05-05 14:37:41 +00:00
|
|
|
* @a: a #ClutterActor
|
2007-11-23 13:11:10 +00:00
|
|
|
*
|
|
|
|
* Evaluates to %TRUE if the %CLUTTER_ACTOR_REALIZED flag is set.
|
|
|
|
*
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* Whether GL resources such as textures are allocated;
|
|
|
|
* if an actor is mapped it must also be realized, but an actor
|
|
|
|
* can be realized and unmapped (this is so hiding an actor temporarily
|
|
|
|
* doesn't do an expensive unrealize/realize).
|
|
|
|
*
|
|
|
|
* To be realized an actor must be inside a stage, and all its parents
|
|
|
|
* must be realized. The stage is required so the actor knows the
|
|
|
|
* correct GL context and window system resources to use.
|
|
|
|
*
|
2007-11-23 13:11:10 +00:00
|
|
|
* Since: 0.2
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
|
|
|
* CLUTTER_ACTOR_IS_VISIBLE:
|
2009-05-05 14:37:41 +00:00
|
|
|
* @a: a #ClutterActor
|
2007-11-23 13:11:10 +00:00
|
|
|
*
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* Evaluates to %TRUE if the actor has been shown, %FALSE if it's hidden.
|
|
|
|
* Equivalent to the ClutterActor::visible object property.
|
|
|
|
*
|
|
|
|
* Note that an actor is only painted onscreen if it's mapped, which
|
|
|
|
* means it's visible, and all its parents are visible, and one of the
|
|
|
|
* parents is a toplevel stage.
|
2007-11-23 13:11:10 +00:00
|
|
|
*
|
|
|
|
* Since: 0.2
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
|
|
|
* CLUTTER_ACTOR_IS_REACTIVE:
|
2009-05-05 14:37:41 +00:00
|
|
|
* @a: a #ClutterActor
|
2007-11-23 13:11:10 +00:00
|
|
|
*
|
|
|
|
* Evaluates to %TRUE if the %CLUTTER_ACTOR_REACTIVE flag is set.
|
|
|
|
*
|
2009-05-05 14:37:41 +00:00
|
|
|
* Only reactive actors will receive event-related signals.
|
|
|
|
*
|
2007-11-23 13:11:10 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
|
2007-10-12 08:17:00 +00:00
|
|
|
#ifdef HAVE_CONFIG_H
|
2006-06-13 13:17:45 +00:00
|
|
|
#include "config.h"
|
2007-10-12 08:17:00 +00:00
|
|
|
#endif
|
2006-06-13 13:17:45 +00:00
|
|
|
|
|
|
|
#include "clutter-actor.h"
|
2007-06-07 14:41:35 +00:00
|
|
|
#include "clutter-container.h"
|
2006-06-13 13:17:45 +00:00
|
|
|
#include "clutter-main.h"
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
#include "clutter-enum-types.h"
|
2007-10-25 14:36:37 +00:00
|
|
|
#include "clutter-scriptable.h"
|
|
|
|
#include "clutter-script.h"
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
#include "clutter-marshal.h"
|
2006-06-13 13:17:45 +00:00
|
|
|
#include "clutter-private.h"
|
2006-12-04 16:26:35 +00:00
|
|
|
#include "clutter-debug.h"
|
2007-05-22 09:31:40 +00:00
|
|
|
#include "clutter-units.h"
|
2009-03-30 15:41:02 +00:00
|
|
|
#include "clutter-profile.h"
|
2008-04-25 13:37:36 +00:00
|
|
|
#include "cogl/cogl.h"
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-12-24 12:53:04 +00:00
|
|
|
typedef struct _ShaderData ShaderData;
|
2009-01-22 13:07:33 +00:00
|
|
|
typedef struct _AnchorCoord AnchorCoord;
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
#define CLUTTER_ACTOR_GET_PRIVATE(obj) \
|
|
|
|
(G_TYPE_INSTANCE_GET_PRIVATE ((obj), CLUTTER_TYPE_ACTOR, ClutterActorPrivate))
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
/* Internal helper struct to represent a point that can be stored in
|
|
|
|
either direct pixel coordinates or as a fraction of the actor's
|
|
|
|
size. It is used for the anchor point, scale center and rotation
|
|
|
|
centers. */
|
|
|
|
struct _AnchorCoord
|
|
|
|
{
|
|
|
|
gboolean is_fractional;
|
|
|
|
|
|
|
|
union
|
|
|
|
{
|
|
|
|
/* Used when is_fractional == TRUE */
|
|
|
|
struct
|
|
|
|
{
|
|
|
|
gdouble x;
|
|
|
|
gdouble y;
|
|
|
|
} fraction;
|
|
|
|
|
|
|
|
/* Use when is_fractional == FALSE */
|
|
|
|
ClutterVertex units;
|
|
|
|
} v;
|
|
|
|
};
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
/* 3 entries should be a good compromise, few layout managers
|
|
|
|
* will ask for 3 different preferred size in each allocation cycle */
|
|
|
|
#define N_CACHED_SIZE_REQUESTS 3
|
|
|
|
typedef struct _SizeRequest SizeRequest;
|
|
|
|
struct _SizeRequest
|
|
|
|
{
|
|
|
|
guint age;
|
|
|
|
gfloat for_size;
|
|
|
|
gfloat min_size;
|
|
|
|
gfloat natural_size;
|
|
|
|
};
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* Internal enum used to control mapped state update. This is a hint
|
|
|
|
* which indicates when to do something other than just enforce
|
|
|
|
* invariants.
|
|
|
|
*/
|
|
|
|
typedef enum {
|
|
|
|
MAP_STATE_CHECK, /* just enforce invariants. */
|
|
|
|
MAP_STATE_MAKE_UNREALIZED, /* force unrealize, ignoring invariants,
|
|
|
|
* used when about to unparent.
|
|
|
|
*/
|
|
|
|
MAP_STATE_MAKE_MAPPED, /* set mapped, error if invariants not met;
|
|
|
|
* used to set mapped on toplevels.
|
|
|
|
*/
|
|
|
|
MAP_STATE_MAKE_UNMAPPED /* set unmapped, even if parent is mapped,
|
|
|
|
* used just before unmapping parent.
|
|
|
|
*/
|
|
|
|
} MapStateChange;
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
struct _ClutterActorPrivate
|
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/* fixed_x, fixed_y, and the allocation box are all in parent
|
|
|
|
* coordinates.
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat fixed_x;
|
|
|
|
gfloat fixed_y;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
/* request mode */
|
|
|
|
ClutterRequestMode request_mode;
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
/* our cached size requests for different width / height */
|
|
|
|
SizeRequest width_requests[N_CACHED_SIZE_REQUESTS];
|
|
|
|
SizeRequest height_requests[N_CACHED_SIZE_REQUESTS];
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
/* An age of 0 means the entry is not set */
|
|
|
|
guint cached_height_age;
|
|
|
|
guint cached_width_age;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
gfloat request_min_width;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat request_min_height;
|
2009-11-12 01:40:57 +00:00
|
|
|
gfloat request_natural_width;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat request_natural_height;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
ClutterActorBox allocation;
|
2009-06-04 11:00:58 +00:00
|
|
|
ClutterAllocationFlags allocation_flags;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
guint position_set : 1;
|
|
|
|
guint min_width_set : 1;
|
|
|
|
guint min_height_set : 1;
|
|
|
|
guint natural_width_set : 1;
|
|
|
|
guint natural_height_set : 1;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/* cached request is invalid (implies allocation is too) */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
guint needs_width_request : 1;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/* cached request is invalid (implies allocation is too) */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
guint needs_height_request : 1;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/* cached allocation is invalid (request has changed, probably) */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
guint needs_allocation : 1;
|
|
|
|
guint queued_redraw : 1;
|
|
|
|
guint show_on_set_parent : 1;
|
|
|
|
guint has_clip : 1;
|
|
|
|
guint clip_to_allocation : 1;
|
|
|
|
guint enable_model_view_transform : 1;
|
|
|
|
guint enable_paint_unmapped : 1;
|
2008-12-23 14:06:55 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat clip[4];
|
2007-12-24 12:53:04 +00:00
|
|
|
|
|
|
|
/* Rotation angles */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gdouble rxang;
|
|
|
|
gdouble ryang;
|
|
|
|
gdouble rzang;
|
2007-12-24 12:53:04 +00:00
|
|
|
|
|
|
|
/* Rotation center: X axis */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
AnchorCoord rx_center;
|
2007-12-24 12:53:04 +00:00
|
|
|
|
|
|
|
/* Rotation center: Y axis */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
AnchorCoord ry_center;
|
2007-12-24 12:53:04 +00:00
|
|
|
|
|
|
|
/* Rotation center: Z axis */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
AnchorCoord rz_center;
|
2007-12-24 12:53:04 +00:00
|
|
|
|
|
|
|
/* Anchor point coordinates */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
AnchorCoord anchor;
|
2007-12-24 12:53:04 +00:00
|
|
|
|
|
|
|
/* depth */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat z;
|
2008-01-30 12:13:26 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
guint8 opacity;
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2007-04-27 21:13:06 +00:00
|
|
|
ClutterActor *parent_actor;
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2007-04-27 21:13:06 +00:00
|
|
|
gchar *name;
|
|
|
|
guint32 id; /* Unique ID */
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
gdouble scale_x;
|
|
|
|
gdouble scale_y;
|
|
|
|
|
|
|
|
AnchorCoord scale_center;
|
2007-12-03 16:29:18 +00:00
|
|
|
|
|
|
|
ShaderData *shader_data;
|
2008-03-28 22:50:55 +00:00
|
|
|
|
2008-12-23 14:06:55 +00:00
|
|
|
PangoContext *pango_context;
|
2009-01-17 16:51:03 +00:00
|
|
|
|
|
|
|
ClutterActor *opacity_parent;
|
2009-11-06 16:55:46 +00:00
|
|
|
|
|
|
|
ClutterTextDirection text_direction;
|
2006-06-13 13:17:45 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
enum
|
|
|
|
{
|
|
|
|
PROP_0,
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
PROP_NAME,
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/* X, Y, WIDTH, HEIGHT are "do what I mean" properties;
|
|
|
|
* when set they force a size request, when gotten they
|
|
|
|
* get the allocation if the allocation is valid, and the
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* request otherwise
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*/
|
2006-06-13 13:17:45 +00:00
|
|
|
PROP_X,
|
|
|
|
PROP_Y,
|
|
|
|
PROP_WIDTH,
|
|
|
|
PROP_HEIGHT,
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
/* Then the rest of these size-related properties are the "actual"
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* underlying properties set or gotten by X, Y, WIDTH, HEIGHT
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*/
|
|
|
|
PROP_FIXED_X,
|
|
|
|
PROP_FIXED_Y,
|
|
|
|
|
|
|
|
PROP_FIXED_POSITION_SET,
|
|
|
|
|
|
|
|
PROP_MIN_WIDTH,
|
|
|
|
PROP_MIN_WIDTH_SET,
|
|
|
|
|
|
|
|
PROP_MIN_HEIGHT,
|
|
|
|
PROP_MIN_HEIGHT_SET,
|
|
|
|
|
|
|
|
PROP_NATURAL_WIDTH,
|
|
|
|
PROP_NATURAL_WIDTH_SET,
|
|
|
|
|
|
|
|
PROP_NATURAL_HEIGHT,
|
|
|
|
PROP_NATURAL_HEIGHT_SET,
|
|
|
|
|
|
|
|
PROP_REQUEST_MODE,
|
|
|
|
|
|
|
|
/* Allocation properties are read-only */
|
|
|
|
PROP_ALLOCATION,
|
|
|
|
|
2007-10-12 12:42:54 +00:00
|
|
|
PROP_DEPTH,
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
PROP_CLIP,
|
|
|
|
PROP_HAS_CLIP,
|
2009-03-16 11:58:58 +00:00
|
|
|
PROP_CLIP_TO_ALLOCATION,
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
PROP_OPACITY,
|
2009-05-05 14:36:29 +00:00
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
PROP_VISIBLE,
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
PROP_MAPPED,
|
2009-05-05 14:36:29 +00:00
|
|
|
PROP_REALIZED,
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
PROP_REACTIVE,
|
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
PROP_SCALE_X,
|
2007-10-17 16:43:46 +00:00
|
|
|
PROP_SCALE_Y,
|
2009-01-23 11:11:24 +00:00
|
|
|
PROP_SCALE_CENTER_X,
|
|
|
|
PROP_SCALE_CENTER_Y,
|
|
|
|
PROP_SCALE_GRAVITY,
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
|
|
|
PROP_ROTATION_ANGLE_X,
|
|
|
|
PROP_ROTATION_ANGLE_Y,
|
|
|
|
PROP_ROTATION_ANGLE_Z,
|
|
|
|
PROP_ROTATION_CENTER_X,
|
|
|
|
PROP_ROTATION_CENTER_Y,
|
2008-03-06 04:37:00 +00:00
|
|
|
PROP_ROTATION_CENTER_Z,
|
2009-01-23 15:55:41 +00:00
|
|
|
/* This property only makes sense for the z rotation because the
|
|
|
|
others would depend on the actor having a size along the
|
|
|
|
z-axis */
|
|
|
|
PROP_ROTATION_CENTER_Z_GRAVITY,
|
2008-03-06 04:37:00 +00:00
|
|
|
|
|
|
|
PROP_ANCHOR_X,
|
2008-04-04 13:20:02 +00:00
|
|
|
PROP_ANCHOR_Y,
|
2009-01-22 13:07:33 +00:00
|
|
|
PROP_ANCHOR_GRAVITY,
|
2008-04-04 13:20:02 +00:00
|
|
|
|
2009-11-06 16:55:46 +00:00
|
|
|
PROP_SHOW_ON_SET_PARENT,
|
|
|
|
|
|
|
|
PROP_TEXT_DIRECTION
|
2006-06-13 13:17:45 +00:00
|
|
|
};
|
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
enum
|
|
|
|
{
|
|
|
|
SHOW,
|
|
|
|
HIDE,
|
|
|
|
DESTROY,
|
2006-11-20 12:02:24 +00:00
|
|
|
PARENT_SET,
|
2009-05-06 16:56:40 +00:00
|
|
|
KEY_FOCUS_IN,
|
|
|
|
KEY_FOCUS_OUT,
|
2008-06-05 14:32:18 +00:00
|
|
|
PAINT,
|
2009-01-23 11:34:13 +00:00
|
|
|
PICK,
|
2008-06-05 14:32:18 +00:00
|
|
|
REALIZE,
|
|
|
|
UNREALIZE,
|
2009-02-17 17:22:02 +00:00
|
|
|
QUEUE_REDRAW,
|
2009-08-08 01:16:04 +00:00
|
|
|
QUEUE_RELAYOUT,
|
2007-08-13 20:48:01 +00:00
|
|
|
EVENT,
|
2007-11-07 16:40:30 +00:00
|
|
|
CAPTURED_EVENT,
|
2007-08-13 20:48:01 +00:00
|
|
|
BUTTON_PRESS_EVENT,
|
|
|
|
BUTTON_RELEASE_EVENT,
|
|
|
|
SCROLL_EVENT,
|
|
|
|
KEY_PRESS_EVENT,
|
|
|
|
KEY_RELEASE_EVENT,
|
|
|
|
MOTION_EVENT,
|
2007-10-09 23:45:49 +00:00
|
|
|
ENTER_EVENT,
|
|
|
|
LEAVE_EVENT,
|
2009-06-04 11:00:58 +00:00
|
|
|
ALLOCATION_CHANGED,
|
2007-11-07 16:40:30 +00:00
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
LAST_SIGNAL
|
|
|
|
};
|
|
|
|
|
|
|
|
static guint actor_signals[LAST_SIGNAL] = { 0, };
|
|
|
|
|
2007-11-30 09:27:10 +00:00
|
|
|
static void clutter_scriptable_iface_init (ClutterScriptableIface *iface);
|
2007-09-27 21:38:38 +00:00
|
|
|
|
2007-11-30 09:27:10 +00:00
|
|
|
static void _clutter_actor_apply_modelview_transform (ClutterActor *self);
|
|
|
|
|
2008-02-18 09:43:01 +00:00
|
|
|
static void clutter_actor_shader_pre_paint (ClutterActor *actor,
|
|
|
|
gboolean repeat);
|
2007-12-03 16:29:18 +00:00
|
|
|
static void clutter_actor_shader_post_paint (ClutterActor *actor);
|
2008-02-18 09:43:01 +00:00
|
|
|
|
2007-12-03 16:29:18 +00:00
|
|
|
static void destroy_shader_data (ClutterActor *self);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/* These setters are all static for now, maybe they should be in the
|
|
|
|
* public API, but they are perhaps obscure enough to leave only as
|
|
|
|
* properties
|
|
|
|
*/
|
|
|
|
static void clutter_actor_set_min_width (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat min_width);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
static void clutter_actor_set_min_height (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat min_height);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
static void clutter_actor_set_natural_width (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat natural_width);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
static void clutter_actor_set_natural_height (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat natural_height);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
static void clutter_actor_set_min_width_set (ClutterActor *self,
|
|
|
|
gboolean use_min_width);
|
|
|
|
static void clutter_actor_set_min_height_set (ClutterActor *self,
|
|
|
|
gboolean use_min_height);
|
|
|
|
static void clutter_actor_set_natural_width_set (ClutterActor *self,
|
|
|
|
gboolean use_natural_width);
|
|
|
|
static void clutter_actor_set_natural_height_set (ClutterActor *self,
|
|
|
|
gboolean use_natural_height);
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
static void clutter_actor_update_map_state (ClutterActor *self,
|
|
|
|
MapStateChange change);
|
|
|
|
static void clutter_actor_unrealize_not_hiding (ClutterActor *self);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
/* Helper routines for managing anchor coords */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
static void clutter_anchor_coord_get_units (ClutterActor *self,
|
2009-01-22 13:07:33 +00:00
|
|
|
const AnchorCoord *coord,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat *x,
|
|
|
|
gfloat *y,
|
|
|
|
gfloat *z);
|
|
|
|
static void clutter_anchor_coord_set_units (AnchorCoord *coord,
|
|
|
|
gfloat x,
|
|
|
|
gfloat y,
|
|
|
|
gfloat z);
|
|
|
|
|
|
|
|
static ClutterGravity clutter_anchor_coord_get_gravity (AnchorCoord *coord);
|
|
|
|
static void clutter_anchor_coord_set_gravity (AnchorCoord *coord,
|
|
|
|
ClutterGravity gravity);
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
static gboolean clutter_anchor_coord_is_zero (const AnchorCoord *coord);
|
|
|
|
|
|
|
|
/* Helper macro which translates by the anchor coord, applies the
|
|
|
|
given transformation and then translates back */
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
#define TRANSFORM_ABOUT_ANCHOR_COORD(a,m,c,_transform) G_STMT_START { \
|
|
|
|
gfloat _tx, _ty, _tz; \
|
|
|
|
clutter_anchor_coord_get_units ((a), (c), &_tx, &_ty, &_tz); \
|
|
|
|
cogl_matrix_translate ((m), _tx, _ty, _tz); \
|
|
|
|
{ _transform; } \
|
|
|
|
cogl_matrix_translate ((m), -_tx, -_ty, -_tz); } G_STMT_END
|
2009-01-22 13:07:33 +00:00
|
|
|
|
2007-11-30 09:27:10 +00:00
|
|
|
G_DEFINE_ABSTRACT_TYPE_WITH_CODE (ClutterActor,
|
|
|
|
clutter_actor,
|
|
|
|
G_TYPE_INITIALLY_UNOWNED,
|
|
|
|
G_IMPLEMENT_INTERFACE (CLUTTER_TYPE_SCRIPTABLE,
|
|
|
|
clutter_scriptable_iface_init));
|
2007-09-27 21:38:38 +00:00
|
|
|
|
2008-02-29 14:54:28 +00:00
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
#ifdef CLUTTER_ENABLE_DEBUG
|
|
|
|
/* XXX - this is for debugging only, remove once working (or leave
|
|
|
|
* in only in some debug mode). Should leave it for a little while
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* until we're confident in the new map/realize/visible handling.
|
|
|
|
*/
|
2009-05-01 11:33:42 +00:00
|
|
|
static inline void
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
clutter_actor_verify_map_state (ClutterActor *self)
|
2006-12-12 20:20:04 +00:00
|
|
|
{
|
2009-05-01 11:33:42 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
if (CLUTTER_ACTOR_IS_REALIZED (self))
|
|
|
|
{
|
|
|
|
/* all bets are off during reparent when we're potentially realized,
|
|
|
|
* but should not be according to invariants
|
|
|
|
*/
|
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_REPARENT))
|
|
|
|
{
|
2009-05-01 11:33:42 +00:00
|
|
|
if (priv->parent_actor == NULL)
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
{
|
|
|
|
if (CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL)
|
|
|
|
{
|
|
|
|
}
|
|
|
|
else
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Realized non-toplevel actor '%s' should "
|
|
|
|
"have a parent",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
2009-05-01 11:33:42 +00:00
|
|
|
else if (!CLUTTER_ACTOR_IS_REALIZED (priv->parent_actor))
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
{
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Realized actor %s has an unrealized parent %s",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self),
|
|
|
|
clutter_actor_get_name (priv->parent_actor)
|
|
|
|
? clutter_actor_get_name (priv->parent_actor)
|
|
|
|
: G_OBJECT_TYPE_NAME (priv->parent_actor));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (CLUTTER_ACTOR_IS_MAPPED (self))
|
2006-12-12 20:20:04 +00:00
|
|
|
{
|
|
|
|
if (!CLUTTER_ACTOR_IS_REALIZED (self))
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Actor '%s' is mapped but not realized",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
2006-12-12 20:20:04 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* remaining bets are off during reparent when we're potentially
|
|
|
|
* mapped, but should not be according to invariants
|
2007-11-26 12:07:25 +00:00
|
|
|
*/
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_REPARENT))
|
|
|
|
{
|
2009-05-01 11:33:42 +00:00
|
|
|
if (priv->parent_actor == NULL)
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
{
|
|
|
|
if (CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL)
|
|
|
|
{
|
2009-05-08 16:14:49 +00:00
|
|
|
if (!CLUTTER_ACTOR_IS_VISIBLE (self) &&
|
|
|
|
!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_DESTRUCTION))
|
|
|
|
{
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Toplevel actor '%s' is mapped "
|
|
|
|
"but not visible",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
2009-05-08 16:14:49 +00:00
|
|
|
}
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Mapped actor '%s' should have a parent",
|
|
|
|
priv->name ? priv->name
|
|
|
|
:G_OBJECT_TYPE_NAME (self));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2009-06-06 15:56:08 +00:00
|
|
|
ClutterActor *iter = self;
|
2009-05-01 11:33:42 +00:00
|
|
|
|
2009-06-06 15:56:08 +00:00
|
|
|
/* check for the enable_paint_unmapped flag on the actor
|
|
|
|
* and parents; if the flag is enabled at any point of this
|
2009-05-01 11:33:42 +00:00
|
|
|
* branch of the scene graph then all the later checks
|
|
|
|
* become pointless
|
|
|
|
*/
|
2009-06-06 15:56:08 +00:00
|
|
|
while (iter != NULL)
|
2009-05-01 11:33:42 +00:00
|
|
|
{
|
2009-06-06 15:56:08 +00:00
|
|
|
if (iter->priv->enable_paint_unmapped)
|
2009-05-01 11:33:42 +00:00
|
|
|
return;
|
|
|
|
|
2009-06-06 15:56:08 +00:00
|
|
|
iter = iter->priv->parent_actor;
|
2009-05-01 11:33:42 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if (!CLUTTER_ACTOR_IS_VISIBLE (priv->parent_actor))
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
{
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Actor '%s' should not be mapped if parent "
|
|
|
|
"is not visible",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
if (!CLUTTER_ACTOR_IS_REALIZED (priv->parent_actor))
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
{
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Actor '%s' should not be mapped if parent "
|
|
|
|
"is not realized",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (priv->parent_actor) &
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
CLUTTER_ACTOR_IS_TOPLEVEL))
|
|
|
|
{
|
2009-05-01 11:33:42 +00:00
|
|
|
if (!CLUTTER_ACTOR_IS_MAPPED (priv->parent_actor))
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Actor '%s' is mapped but its non-toplevel "
|
|
|
|
"parent is not mapped",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
#endif /* CLUTTER_ENABLE_DEBUG */
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_set_mapped (ClutterActor *self,
|
|
|
|
gboolean mapped)
|
|
|
|
{
|
|
|
|
if (CLUTTER_ACTOR_IS_MAPPED (self) == mapped)
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (mapped)
|
|
|
|
{
|
|
|
|
CLUTTER_ACTOR_GET_CLASS (self)->map (self);
|
|
|
|
g_assert (CLUTTER_ACTOR_IS_MAPPED (self));
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
CLUTTER_ACTOR_GET_CLASS (self)->unmap (self);
|
|
|
|
g_assert (!CLUTTER_ACTOR_IS_MAPPED (self));
|
|
|
|
}
|
|
|
|
}
|
2006-12-12 20:20:04 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* this function updates the mapped and realized states according to
|
|
|
|
* invariants, in the appropriate order.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
clutter_actor_update_map_state (ClutterActor *self,
|
|
|
|
MapStateChange change)
|
|
|
|
{
|
|
|
|
gboolean was_mapped;
|
|
|
|
gboolean was_realized;
|
|
|
|
|
|
|
|
was_mapped = CLUTTER_ACTOR_IS_MAPPED (self);
|
|
|
|
was_realized = CLUTTER_ACTOR_IS_REALIZED (self);
|
|
|
|
|
|
|
|
if (CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL)
|
|
|
|
{
|
|
|
|
/* the mapped flag on top-level actors must be set by the
|
|
|
|
* per-backend implementation because it might be asynchronous.
|
|
|
|
*
|
|
|
|
* That is, the MAPPED flag on toplevels currently tracks the X
|
|
|
|
* server mapped-ness of the window, while the expected behavior
|
|
|
|
* (if used to GTK) may be to track WM_STATE!=WithdrawnState.
|
|
|
|
* This creates some weird complexity by breaking the invariant
|
|
|
|
* that if we're visible and all ancestors shown then we are
|
|
|
|
* also mapped - instead, we are mapped if all ancestors
|
|
|
|
* _possibly excepting_ the stage are mapped. The stage
|
|
|
|
* will map/unmap for example when it is minimized or
|
|
|
|
* moved to another workspace.
|
|
|
|
*
|
|
|
|
* So, the only invariant on the stage is that if visible it
|
|
|
|
* should be realized, and that it has to be visible to be
|
|
|
|
* mapped.
|
|
|
|
*/
|
2006-12-12 20:20:04 +00:00
|
|
|
if (CLUTTER_ACTOR_IS_VISIBLE (self))
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
clutter_actor_realize (self);
|
|
|
|
|
|
|
|
switch (change)
|
|
|
|
{
|
|
|
|
case MAP_STATE_CHECK:
|
|
|
|
break;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
case MAP_STATE_MAKE_MAPPED:
|
|
|
|
g_assert (!was_mapped);
|
|
|
|
clutter_actor_set_mapped (self, TRUE);
|
|
|
|
break;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
case MAP_STATE_MAKE_UNMAPPED:
|
|
|
|
g_assert (was_mapped);
|
|
|
|
clutter_actor_set_mapped (self, FALSE);
|
|
|
|
break;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
case MAP_STATE_MAKE_UNREALIZED:
|
|
|
|
/* we only use MAKE_UNREALIZED in unparent,
|
|
|
|
* and unparenting a stage isn't possible.
|
|
|
|
* If someone wants to just unrealize a stage
|
|
|
|
* then clutter_actor_unrealize() doesn't
|
|
|
|
* go through this codepath.
|
|
|
|
*/
|
|
|
|
g_warning ("Trying to force unrealize stage is not allowed");
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2009-05-08 16:14:49 +00:00
|
|
|
if (CLUTTER_ACTOR_IS_MAPPED (self) &&
|
|
|
|
!CLUTTER_ACTOR_IS_VISIBLE (self) &&
|
|
|
|
!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_DESTRUCTION))
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
{
|
2009-06-04 15:27:21 +00:00
|
|
|
g_warning ("Clutter toplevel of type '%s' is not visible, but "
|
|
|
|
"it is somehow still mapped",
|
|
|
|
G_OBJECT_TYPE_NAME (self));
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
}
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
|
|
|
ClutterActor *parent = priv->parent_actor;
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
gboolean should_be_mapped;
|
|
|
|
gboolean may_be_realized;
|
|
|
|
gboolean must_be_realized;
|
|
|
|
|
|
|
|
should_be_mapped = FALSE;
|
|
|
|
may_be_realized = TRUE;
|
|
|
|
must_be_realized = FALSE;
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (parent == NULL || change == MAP_STATE_MAKE_UNREALIZED)
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
{
|
|
|
|
may_be_realized = FALSE;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
/* Maintain invariant that if parent is mapped, and we are
|
|
|
|
* visible, then we are mapped ... unless parent is a
|
|
|
|
* stage, in which case we map regardless of parent's map
|
|
|
|
* state but do require stage to be visible and realized.
|
|
|
|
*
|
|
|
|
* If parent is realized, that does not force us to be
|
|
|
|
* realized; but if parent is unrealized, that does force
|
|
|
|
* us to be unrealized.
|
|
|
|
*
|
|
|
|
* The reason we don't force children to realize with
|
|
|
|
* parents is _clutter_actor_rerealize(); if we require that
|
|
|
|
* a realized parent means children are realized, then to
|
|
|
|
* unrealize an actor we would have to unrealize its
|
|
|
|
* parents, which would end up meaning unrealizing and
|
|
|
|
* hiding the entire stage. So we allow unrealizing a
|
|
|
|
* child (as long as that child is not mapped) while that
|
|
|
|
* child still has a realized parent.
|
|
|
|
*
|
|
|
|
* Also, if we unrealize from leaf nodes to root, and
|
|
|
|
* realize from root to leaf, the invariants are never
|
|
|
|
* violated if we allow children to be unrealized
|
|
|
|
* while parents are realized.
|
|
|
|
*
|
|
|
|
* When unmapping, MAP_STATE_MAKE_UNMAPPED is specified
|
|
|
|
* to force us to unmap, even though parent is still
|
|
|
|
* mapped. This is because we're unmapping from leaf nodes
|
|
|
|
* up to root nodes.
|
|
|
|
*/
|
|
|
|
if (CLUTTER_ACTOR_IS_VISIBLE (self) &&
|
|
|
|
change != MAP_STATE_MAKE_UNMAPPED)
|
|
|
|
{
|
|
|
|
gboolean parent_is_visible_realized_toplevel;
|
|
|
|
|
|
|
|
parent_is_visible_realized_toplevel =
|
|
|
|
(((CLUTTER_PRIVATE_FLAGS (parent) &
|
|
|
|
CLUTTER_ACTOR_IS_TOPLEVEL) != 0) &&
|
|
|
|
CLUTTER_ACTOR_IS_VISIBLE (parent) &&
|
|
|
|
CLUTTER_ACTOR_IS_REALIZED (parent));
|
|
|
|
|
|
|
|
if (CLUTTER_ACTOR_IS_MAPPED (parent) ||
|
|
|
|
parent_is_visible_realized_toplevel)
|
|
|
|
{
|
|
|
|
must_be_realized = TRUE;
|
|
|
|
should_be_mapped = TRUE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
/* if the actor has been set to be painted even if unmapped
|
|
|
|
* then we should map it and check for realization as well;
|
|
|
|
* this is an override for the branch of the scene graph
|
|
|
|
* which begins with this node
|
|
|
|
*/
|
|
|
|
if (priv->enable_paint_unmapped)
|
|
|
|
{
|
|
|
|
if (priv->parent_actor == NULL)
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Attempting to map an unparented actor '%s'",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
2009-05-01 11:33:42 +00:00
|
|
|
|
|
|
|
should_be_mapped = TRUE;
|
|
|
|
must_be_realized = TRUE;
|
|
|
|
}
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
if (!CLUTTER_ACTOR_IS_REALIZED (parent))
|
|
|
|
may_be_realized = FALSE;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (change == MAP_STATE_MAKE_MAPPED && !should_be_mapped)
|
|
|
|
{
|
2009-08-04 10:11:59 +00:00
|
|
|
if (parent == NULL)
|
|
|
|
g_warning ("Attempting to map a child that does not "
|
2009-10-23 08:48:35 +00:00
|
|
|
"meet the necessary invariants: the actor '%s' "
|
|
|
|
"has no parent",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
2009-08-04 10:11:59 +00:00
|
|
|
else
|
|
|
|
g_warning ("Attempting to map a child that does not "
|
2009-10-23 08:48:35 +00:00
|
|
|
"meet the necessary invariants: the actor '%s' "
|
|
|
|
"is parented to an unmapped actor",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* If in reparent, we temporarily suspend unmap and unrealize.
|
|
|
|
*
|
|
|
|
* We want to go in the order "realize, map" and "unmap, unrealize"
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* Unmap */
|
|
|
|
if (!should_be_mapped &&
|
|
|
|
!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_REPARENT))
|
|
|
|
{
|
|
|
|
clutter_actor_set_mapped (self, FALSE);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Realize */
|
|
|
|
if (must_be_realized)
|
|
|
|
clutter_actor_realize (self);
|
|
|
|
|
|
|
|
/* if we must be realized then we may be, presumably */
|
|
|
|
g_assert (!(must_be_realized && !may_be_realized));
|
|
|
|
|
|
|
|
/* Unrealize */
|
|
|
|
if (!may_be_realized &&
|
|
|
|
!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_REPARENT))
|
|
|
|
clutter_actor_unrealize_not_hiding (self);
|
|
|
|
|
|
|
|
/* Map */
|
|
|
|
if (should_be_mapped)
|
|
|
|
{
|
|
|
|
if (!must_be_realized)
|
2009-10-23 08:48:35 +00:00
|
|
|
g_warning ("Somehow we think actor '%s' should be mapped but "
|
|
|
|
"not realized, which isn't allowed",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
|
|
|
/* realization is allowed to fail (though I don't know what
|
|
|
|
* an app is supposed to do about that - shouldn't it just
|
|
|
|
* be a g_error? anyway, we have to avoid mapping if this
|
|
|
|
* happens)
|
|
|
|
*/
|
|
|
|
if (CLUTTER_ACTOR_IS_REALIZED (self))
|
2009-05-01 11:33:42 +00:00
|
|
|
clutter_actor_set_mapped (self, TRUE);
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-05-11 11:43:56 +00:00
|
|
|
#ifdef CLUTTER_ENABLE_DEBUG
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* check all invariants were kept */
|
|
|
|
clutter_actor_verify_map_state (self);
|
2009-05-11 11:43:56 +00:00
|
|
|
#endif
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_real_map (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_assert (!CLUTTER_ACTOR_IS_MAPPED (self));
|
|
|
|
|
|
|
|
CLUTTER_ACTOR_SET_FLAGS (self, CLUTTER_ACTOR_MAPPED);
|
2009-10-07 15:41:25 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* notify on parent mapped before potentially mapping
|
|
|
|
* children, so apps see a top-down notification.
|
|
|
|
*/
|
|
|
|
g_object_notify (G_OBJECT (self), "mapped");
|
|
|
|
|
|
|
|
if (CLUTTER_IS_CONTAINER (self))
|
2009-05-01 11:31:06 +00:00
|
|
|
clutter_container_foreach_with_internals (CLUTTER_CONTAINER (self),
|
|
|
|
CLUTTER_CALLBACK (clutter_actor_map),
|
|
|
|
NULL);
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_map:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2009-05-01 11:31:06 +00:00
|
|
|
* Sets the #CLUTTER_ACTOR_MAPPED flag on the actor and possibly maps
|
|
|
|
* and realizes its children if they are visible. Does nothing if the
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* actor is not visible.
|
|
|
|
*
|
2009-05-01 11:31:06 +00:00
|
|
|
* Calling this is allowed in only one case: you are implementing the
|
|
|
|
* #ClutterActor::map virtual function in an actor and you need to map
|
|
|
|
* the children of that actor. It is not necessary to call this
|
|
|
|
* if you implement #ClutterContainer because the default implementation
|
|
|
|
* will automatically map children of containers.
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
*
|
|
|
|
* When overriding map, it is mandatory to chain up to the parent
|
|
|
|
* implementation.
|
2009-05-01 11:31:06 +00:00
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
void
|
|
|
|
clutter_actor_map (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
if (CLUTTER_ACTOR_IS_MAPPED (self))
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (!CLUTTER_ACTOR_IS_VISIBLE (self))
|
|
|
|
return;
|
|
|
|
|
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_MAKE_MAPPED);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_real_unmap (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_assert (CLUTTER_ACTOR_IS_MAPPED (self));
|
|
|
|
|
|
|
|
if (CLUTTER_IS_CONTAINER (self))
|
2009-05-01 11:31:06 +00:00
|
|
|
clutter_container_foreach_with_internals (CLUTTER_CONTAINER (self),
|
|
|
|
CLUTTER_CALLBACK (clutter_actor_unmap),
|
|
|
|
NULL);
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
|
|
|
CLUTTER_ACTOR_UNSET_FLAGS (self, CLUTTER_ACTOR_MAPPED);
|
2009-05-05 14:37:41 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* notify on parent mapped after potentially unmapping
|
|
|
|
* children, so apps see a bottom-up notification.
|
|
|
|
*/
|
|
|
|
g_object_notify (G_OBJECT (self), "mapped");
|
|
|
|
|
2009-04-08 09:33:39 +00:00
|
|
|
/* relinquish keyboard focus if we were unmapped while owning it */
|
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL))
|
|
|
|
{
|
|
|
|
ClutterActor *stage;
|
|
|
|
|
|
|
|
stage = clutter_actor_get_stage (self);
|
|
|
|
|
|
|
|
if (stage &&
|
|
|
|
clutter_stage_get_key_focus (CLUTTER_STAGE (stage)) == self)
|
|
|
|
{
|
|
|
|
clutter_stage_set_key_focus (CLUTTER_STAGE (stage), NULL);
|
|
|
|
}
|
|
|
|
}
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_unmap:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
|
|
|
* Unsets the #CLUTTER_ACTOR_MAPPED flag on the actor and possibly
|
|
|
|
* unmaps its children if they were mapped.
|
|
|
|
*
|
2009-05-01 11:31:06 +00:00
|
|
|
* Calling this is allowed in only one case: you are implementing the
|
|
|
|
* #ClutterActor::unmap virtual function in an actor and you need to
|
|
|
|
* unmap the children of that actor. It is not necessary to call this
|
|
|
|
* if you implement #ClutterContainer because the default implementation
|
|
|
|
* will automatically unmap children of containers.
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
*
|
|
|
|
* When overriding unmap, it is mandatory to chain up to the parent
|
|
|
|
* implementation.
|
2009-05-01 11:31:06 +00:00
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
void
|
|
|
|
clutter_actor_unmap (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
if (!CLUTTER_ACTOR_IS_MAPPED (self))
|
|
|
|
return;
|
|
|
|
|
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_MAKE_UNMAPPED);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_real_show (ClutterActor *self)
|
|
|
|
{
|
|
|
|
if (!CLUTTER_ACTOR_IS_VISIBLE (self))
|
|
|
|
{
|
2009-10-06 03:20:07 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
CLUTTER_ACTOR_SET_FLAGS (self, CLUTTER_ACTOR_VISIBLE);
|
2009-10-07 15:41:25 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* we notify on the "visible" flag in the clutter_actor_show()
|
|
|
|
* wrapper so the entire show signal emission completes first
|
|
|
|
* (?)
|
|
|
|
*/
|
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_CHECK);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2009-10-07 15:41:25 +00:00
|
|
|
/* we queue a relayout unless the actor is inside a
|
|
|
|
* container that explicitly told us not to
|
2009-10-06 03:20:07 +00:00
|
|
|
*/
|
2009-10-07 15:41:25 +00:00
|
|
|
if (priv->parent_actor &&
|
|
|
|
(!(priv->parent_actor->flags & CLUTTER_ACTOR_NO_LAYOUT)))
|
|
|
|
{
|
|
|
|
/* While an actor is hidden the parent may not have
|
|
|
|
* allocated/requested so we need to start from scratch
|
|
|
|
* and avoid the short-circuiting in
|
|
|
|
* clutter_actor_queue_relayout().
|
|
|
|
*/
|
|
|
|
priv->needs_width_request = FALSE;
|
|
|
|
priv->needs_height_request = FALSE;
|
|
|
|
priv->needs_allocation = FALSE;
|
|
|
|
clutter_actor_queue_relayout (self);
|
|
|
|
}
|
2006-12-12 20:20:04 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
2008-04-15 13:55:36 +00:00
|
|
|
* clutter_actor_show:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Flags an actor to be displayed. An actor that isn't shown will not
|
2008-02-15 14:39:25 +00:00
|
|
|
* be rendered on the stage.
|
2008-04-23 13:54:13 +00:00
|
|
|
*
|
|
|
|
* Actors are visible by default.
|
|
|
|
*
|
|
|
|
* If this function is called on an actor without a parent, the
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* #ClutterActor:show-on-set-parent will be set to %TRUE as a side
|
|
|
|
* effect.
|
2008-02-15 14:39:25 +00:00
|
|
|
*/
|
2006-06-13 13:17:45 +00:00
|
|
|
void
|
|
|
|
clutter_actor_show (ClutterActor *self)
|
|
|
|
{
|
2008-04-04 13:20:02 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
#ifdef CLUTTER_ENABLE_DEBUG
|
|
|
|
clutter_actor_verify_map_state (self);
|
|
|
|
#endif
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
2008-04-04 13:20:02 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
2008-04-23 13:54:13 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
2008-04-04 13:20:02 +00:00
|
|
|
|
2008-04-23 13:54:13 +00:00
|
|
|
if (!priv->show_on_set_parent && !priv->parent_actor)
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
{
|
2008-04-23 13:54:13 +00:00
|
|
|
priv->show_on_set_parent = TRUE;
|
|
|
|
g_object_notify (G_OBJECT (self), "show-on-set-parent");
|
|
|
|
}
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-04-23 13:54:13 +00:00
|
|
|
if (!CLUTTER_ACTOR_IS_VISIBLE (self))
|
|
|
|
{
|
2006-12-12 20:20:04 +00:00
|
|
|
g_signal_emit (self, actor_signals[SHOW], 0);
|
|
|
|
g_object_notify (G_OBJECT (self), "visible");
|
|
|
|
}
|
2008-04-23 13:54:13 +00:00
|
|
|
|
2009-06-11 22:56:27 +00:00
|
|
|
if (priv->parent_actor)
|
|
|
|
clutter_actor_queue_redraw (priv->parent_actor);
|
|
|
|
|
2008-04-23 13:54:13 +00:00
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2006-12-12 20:20:04 +00:00
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2006-12-12 20:20:04 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_show_all:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Calls clutter_actor_show() on all children of an actor (if any).
|
2006-12-12 20:20:04 +00:00
|
|
|
*
|
|
|
|
* Since: 0.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_show_all (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorClass *klass;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2006-12-12 20:20:04 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2006-12-12 20:20:04 +00:00
|
|
|
klass = CLUTTER_ACTOR_GET_CLASS (self);
|
|
|
|
if (klass->show_all)
|
|
|
|
klass->show_all (self);
|
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2006-12-12 20:20:04 +00:00
|
|
|
void
|
|
|
|
clutter_actor_real_hide (ClutterActor *self)
|
|
|
|
{
|
|
|
|
if (CLUTTER_ACTOR_IS_VISIBLE (self))
|
|
|
|
{
|
2009-10-07 15:41:25 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
CLUTTER_ACTOR_UNSET_FLAGS (self, CLUTTER_ACTOR_VISIBLE);
|
2009-05-05 14:37:41 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* we notify on the "visible" flag in the clutter_actor_hide()
|
|
|
|
* wrapper so the entire hide signal emission completes first
|
|
|
|
* (?)
|
2007-11-26 12:07:25 +00:00
|
|
|
*/
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_CHECK);
|
2007-11-26 12:07:25 +00:00
|
|
|
|
2009-10-07 15:41:25 +00:00
|
|
|
/* we queue a relayout unless the actor is inside a
|
|
|
|
* container that explicitly told us not to
|
|
|
|
*/
|
|
|
|
if (priv->parent_actor &&
|
|
|
|
(!(priv->parent_actor->flags & CLUTTER_ACTOR_NO_LAYOUT)))
|
|
|
|
clutter_actor_queue_relayout (priv->parent_actor);
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2008-04-15 13:55:36 +00:00
|
|
|
* clutter_actor_hide:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Flags an actor to be hidden. A hidden actor will not be
|
2008-02-15 14:39:25 +00:00
|
|
|
* rendered on the stage.
|
2008-04-23 13:54:13 +00:00
|
|
|
*
|
|
|
|
* Actors are visible by default.
|
|
|
|
*
|
|
|
|
* If this function is called on an actor without a parent, the
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* #ClutterActor:show-on-set-parent property will be set to %FALSE
|
2008-04-23 13:54:13 +00:00
|
|
|
* as a side-effect.
|
2008-02-15 14:39:25 +00:00
|
|
|
*/
|
2006-06-13 13:17:45 +00:00
|
|
|
void
|
|
|
|
clutter_actor_hide (ClutterActor *self)
|
|
|
|
{
|
2008-04-04 13:20:02 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
#ifdef CLUTTER_ENABLE_DEBUG
|
|
|
|
clutter_actor_verify_map_state (self);
|
|
|
|
#endif
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
2008-04-04 13:20:02 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
2008-04-23 13:54:13 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
2008-04-04 13:20:02 +00:00
|
|
|
|
2008-04-23 13:54:13 +00:00
|
|
|
if (priv->show_on_set_parent && !priv->parent_actor)
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
{
|
2008-04-23 13:54:13 +00:00
|
|
|
priv->show_on_set_parent = FALSE;
|
|
|
|
g_object_notify (G_OBJECT (self), "show-on-set-parent");
|
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
if (CLUTTER_ACTOR_IS_VISIBLE (self))
|
2008-04-23 13:54:13 +00:00
|
|
|
{
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
g_signal_emit (self, actor_signals[HIDE], 0);
|
|
|
|
g_object_notify (G_OBJECT (self), "visible");
|
2007-10-12 08:17:00 +00:00
|
|
|
}
|
2008-04-23 13:54:13 +00:00
|
|
|
|
2009-06-11 22:56:27 +00:00
|
|
|
if (priv->parent_actor)
|
|
|
|
clutter_actor_queue_redraw (priv->parent_actor);
|
|
|
|
|
2008-04-23 13:54:13 +00:00
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
2006-12-12 20:20:04 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_hide_all:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Calls clutter_actor_hide() on all child actors (if any).
|
2006-12-12 20:20:04 +00:00
|
|
|
*
|
|
|
|
* Since: 0.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_hide_all (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorClass *klass;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
klass = CLUTTER_ACTOR_GET_CLASS (self);
|
|
|
|
if (klass->hide_all)
|
|
|
|
klass->hide_all (self);
|
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
2008-04-15 13:55:36 +00:00
|
|
|
* clutter_actor_realize:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
|
|
|
* Creates any underlying graphics resources needed by the actor to be
|
2007-10-12 08:17:00 +00:00
|
|
|
* displayed.
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
*
|
|
|
|
* Realization means the actor is now tied to a specific rendering context
|
|
|
|
* (that is, a specific toplevel stage).
|
|
|
|
*
|
|
|
|
* This function does nothing if the actor is already realized.
|
|
|
|
*
|
|
|
|
* Because a realized actor must have realized parent actors, calling
|
|
|
|
* clutter_actor_realize() will also realize all parents of the actor.
|
|
|
|
*
|
|
|
|
* This function does not realize child actors, except in the special
|
|
|
|
* case that realizing the stage, when the stage is visible, will
|
|
|
|
* suddenly map (and thus realize) the children of the stage.
|
2006-06-13 13:17:45 +00:00
|
|
|
**/
|
|
|
|
void
|
|
|
|
clutter_actor_realize (ClutterActor *self)
|
|
|
|
{
|
2008-06-05 14:32:18 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
#ifdef CLUTTER_ENABLE_DEBUG
|
|
|
|
clutter_actor_verify_map_state (self);
|
|
|
|
#endif
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
if (CLUTTER_ACTOR_IS_REALIZED (self))
|
|
|
|
return;
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* To be realized, our parent actors must be realized first.
|
|
|
|
* This will only succeed if we're inside a toplevel.
|
|
|
|
*/
|
2009-05-08 16:14:49 +00:00
|
|
|
if (self->priv->parent_actor != NULL)
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
clutter_actor_realize (self->priv->parent_actor);
|
|
|
|
|
|
|
|
if (CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL)
|
|
|
|
{
|
|
|
|
/* toplevels can be realized at any time */
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
/* "Fail" the realization if parent is missing or unrealized;
|
|
|
|
* this should really be a g_warning() not some kind of runtime
|
|
|
|
* failure; how can an app possibly recover? Instead it's a bug
|
|
|
|
* in the app and the app should get an explanatory warning so
|
|
|
|
* someone can fix it. But for now it's too hard to fix this
|
|
|
|
* because e.g. ClutterTexture needs reworking.
|
|
|
|
*/
|
|
|
|
if (self->priv->parent_actor == NULL ||
|
|
|
|
!CLUTTER_ACTOR_IS_REALIZED (self->priv->parent_actor))
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2009-05-11 11:36:14 +00:00
|
|
|
CLUTTER_NOTE (ACTOR, "Realizing actor '%s' [%p]",
|
|
|
|
self->priv->name ? self->priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self),
|
|
|
|
self);
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
CLUTTER_ACTOR_SET_FLAGS (self, CLUTTER_ACTOR_REALIZED);
|
2009-05-05 14:36:29 +00:00
|
|
|
g_object_notify (G_OBJECT (self), "realized");
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2008-06-05 14:32:18 +00:00
|
|
|
g_signal_emit (self, actor_signals[REALIZE], 0);
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
|
|
|
/* Stage actor is allowed to unset the realized flag again in its
|
|
|
|
* default signal handler, though that is a pathological situation.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* If realization "failed" we'll have to update child state. */
|
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_CHECK);
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
clutter_actor_real_unrealize (ClutterActor *self)
|
|
|
|
{
|
|
|
|
/* we must be unmapped (implying our children are also unmapped) */
|
|
|
|
g_assert (!CLUTTER_ACTOR_IS_MAPPED (self));
|
|
|
|
|
|
|
|
if (CLUTTER_IS_CONTAINER (self))
|
2009-05-01 11:31:06 +00:00
|
|
|
clutter_container_foreach_with_internals (CLUTTER_CONTAINER (self),
|
|
|
|
CLUTTER_CALLBACK (clutter_actor_unrealize_not_hiding),
|
|
|
|
NULL);
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2008-04-15 13:55:36 +00:00
|
|
|
* clutter_actor_unrealize:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* Frees up any underlying graphics resources needed by the actor to
|
|
|
|
* be displayed.
|
|
|
|
*
|
|
|
|
* Unrealization means the actor is now independent of any specific
|
|
|
|
* rendering context (is not attached to a specific toplevel stage).
|
|
|
|
*
|
|
|
|
* Because mapped actors must be realized, actors may not be
|
|
|
|
* unrealized if they are mapped. This function hides the actor to be
|
|
|
|
* sure it isn't mapped, an application-visible side effect that you
|
|
|
|
* may not be expecting.
|
|
|
|
*
|
|
|
|
* This function should not really be in the public API, because
|
|
|
|
* there isn't a good reason to call it. ClutterActor will already
|
|
|
|
* unrealize things for you when it's important to do so.
|
|
|
|
*
|
|
|
|
* If you were using clutter_actor_unrealize() in a dispose
|
|
|
|
* implementation, then don't, just chain up to ClutterActor's
|
|
|
|
* dispose.
|
|
|
|
*
|
|
|
|
* If you were using clutter_actor_unrealize() to implement
|
|
|
|
* unrealizing children of your container, then don't, ClutterActor
|
|
|
|
* will already take care of that.
|
|
|
|
*
|
|
|
|
* If you were using clutter_actor_unrealize() to re-realize to
|
|
|
|
* create your resources in a different way, then use
|
|
|
|
* _clutter_actor_rerealize() (inside Clutter) or just call your
|
|
|
|
* code that recreates your resources directly (outside Clutter).
|
|
|
|
*/
|
2006-06-13 13:17:45 +00:00
|
|
|
void
|
|
|
|
clutter_actor_unrealize (ClutterActor *self)
|
|
|
|
{
|
2008-06-05 14:32:18 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
g_return_if_fail (!CLUTTER_ACTOR_IS_MAPPED (self));
|
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
#ifdef CLUTTER_ENABLE_DEBUG
|
|
|
|
clutter_actor_verify_map_state (self);
|
|
|
|
#endif
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
|
|
|
clutter_actor_hide (self);
|
|
|
|
|
|
|
|
clutter_actor_unrealize_not_hiding (self);
|
|
|
|
}
|
|
|
|
|
2009-10-23 09:08:02 +00:00
|
|
|
/*
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* clutter_actor_unrealize_not_hiding:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
|
|
|
* Frees up any underlying graphics resources needed by the actor to
|
|
|
|
* be displayed.
|
|
|
|
*
|
|
|
|
* Unrealization means the actor is now independent of any specific
|
|
|
|
* rendering context (is not attached to a specific toplevel stage).
|
|
|
|
*
|
|
|
|
* Because mapped actors must be realized, actors may not be
|
|
|
|
* unrealized if they are mapped. You must hide the actor or one of
|
|
|
|
* its parents before attempting to unrealize.
|
|
|
|
*
|
|
|
|
* This function is separate from clutter_actor_unrealize() because it
|
|
|
|
* does not automatically hide the actor.
|
|
|
|
* Actors need not be hidden to be unrealized, they just need to
|
|
|
|
* be unmapped. In fact we don't want to mess up the application's
|
|
|
|
* setting of the "visible" flag, so hiding is very undesirable.
|
|
|
|
*
|
|
|
|
* clutter_actor_unrealize() does a clutter_actor_hide() just for
|
|
|
|
* backward compatibility.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
clutter_actor_unrealize_not_hiding (ClutterActor *self)
|
|
|
|
{
|
|
|
|
/* All callers of clutter_actor_unrealize_not_hiding() should have
|
|
|
|
* taken care of unmapping the actor first. This means
|
|
|
|
* all our children should also be unmapped.
|
|
|
|
*/
|
|
|
|
g_assert (!CLUTTER_ACTOR_IS_MAPPED (self));
|
2006-06-13 13:17:45 +00:00
|
|
|
|
|
|
|
if (!CLUTTER_ACTOR_IS_REALIZED (self))
|
|
|
|
return;
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* The default handler for the signal should recursively unrealize
|
|
|
|
* child actors. We want to unset the realized flag only _after_
|
|
|
|
* child actors are unrealized, to maintain invariants.
|
2008-06-13 14:16:22 +00:00
|
|
|
*/
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
|
|
|
g_signal_emit (self, actor_signals[UNREALIZE], 0);
|
2008-06-13 14:16:22 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
CLUTTER_ACTOR_UNSET_FLAGS (self, CLUTTER_ACTOR_REALIZED);
|
2009-05-05 14:36:29 +00:00
|
|
|
|
|
|
|
g_object_notify (G_OBJECT (self), "realized");
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2009-10-23 09:08:02 +00:00
|
|
|
/*
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* _clutter_actor_rerealize:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
* @callback: Function to call while unrealized
|
|
|
|
* @data: data for callback
|
|
|
|
*
|
|
|
|
* If an actor is already unrealized, this just calls the callback.
|
|
|
|
*
|
|
|
|
* If it is realized, it unrealizes temporarily, calls the callback,
|
|
|
|
* and then re-realizes the actor.
|
|
|
|
*
|
|
|
|
* As a side effect, leaves all children of the actor unrealized if
|
|
|
|
* the actor was realized but not showing. This is because when we
|
|
|
|
* unrealize the actor temporarily we must unrealize its children
|
|
|
|
* (e.g. children of a stage can't be realized if stage window is
|
|
|
|
* gone). And we aren't clever enough to save the realization state of
|
|
|
|
* all children. In most cases this should not matter, because
|
|
|
|
* the children will automatically realize when they next become mapped.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
_clutter_actor_rerealize (ClutterActor *self,
|
|
|
|
ClutterCallback callback,
|
|
|
|
void *data)
|
|
|
|
{
|
|
|
|
gboolean was_mapped;
|
|
|
|
gboolean was_showing;
|
|
|
|
gboolean was_realized;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
#ifdef CLUTTER_ENABLE_DEBUG
|
|
|
|
clutter_actor_verify_map_state (self);
|
|
|
|
#endif
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
|
|
|
was_realized = CLUTTER_ACTOR_IS_REALIZED (self);
|
|
|
|
was_mapped = CLUTTER_ACTOR_IS_MAPPED (self);
|
|
|
|
was_showing = CLUTTER_ACTOR_IS_VISIBLE (self);
|
|
|
|
|
|
|
|
/* Must be unmapped to unrealize. Note we only have to hide this
|
|
|
|
* actor if it was mapped (if all parents were showing). If actor
|
|
|
|
* is merely visible (but not mapped), then that's fine, we can
|
|
|
|
* leave it visible.
|
|
|
|
*/
|
|
|
|
if (was_mapped)
|
|
|
|
clutter_actor_hide (self);
|
|
|
|
|
|
|
|
g_assert (!CLUTTER_ACTOR_IS_MAPPED (self));
|
|
|
|
|
|
|
|
/* unrealize self and all children */
|
|
|
|
clutter_actor_unrealize_not_hiding (self);
|
|
|
|
|
|
|
|
if (callback != NULL)
|
|
|
|
{
|
|
|
|
(* callback) (self, data);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (was_showing)
|
|
|
|
clutter_actor_show (self); /* will realize only if mapping implies it */
|
|
|
|
else if (was_realized)
|
|
|
|
clutter_actor_realize (self); /* realize self and all parents */
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
2007-11-18 23:24:55 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_real_pick (ClutterActor *self,
|
|
|
|
const ClutterColor *color)
|
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/* the default implementation is just to paint a rectangle
|
|
|
|
* with the same size of the actor using the passed color
|
|
|
|
*/
|
2007-11-18 23:24:55 +00:00
|
|
|
if (clutter_actor_should_pick_paint (self))
|
|
|
|
{
|
2009-01-22 15:55:43 +00:00
|
|
|
ClutterActorBox box = { 0, };
|
|
|
|
float width, height;
|
|
|
|
|
|
|
|
clutter_actor_get_allocation_box (self, &box);
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
width = box.x2 - box.x1;
|
|
|
|
height = box.y2 - box.y1;
|
2009-01-22 15:55:43 +00:00
|
|
|
|
2008-11-12 13:57:58 +00:00
|
|
|
cogl_set_source_color4ub (color->red,
|
|
|
|
color->green,
|
|
|
|
color->blue,
|
|
|
|
color->alpha);
|
2009-01-22 15:55:43 +00:00
|
|
|
|
|
|
|
cogl_rectangle (0, 0, width, height);
|
2007-11-18 23:24:55 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_should_pick_paint:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2009-03-19 11:40:13 +00:00
|
|
|
* Should be called inside the implementation of the
|
|
|
|
* #ClutterActor::pick virtual function in order to check whether
|
|
|
|
* the actor should paint itself in pick mode or not.
|
2007-11-18 23:24:55 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* This function should never be called directly by applications.
|
2007-11-18 23:24:55 +00:00
|
|
|
*
|
2007-12-17 17:45:49 +00:00
|
|
|
* Return value: %TRUE if the actor should paint its silhouette,
|
|
|
|
* %FALSE otherwise
|
2007-11-18 23:24:55 +00:00
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_should_pick_paint (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterMainContext *context;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), FALSE);
|
|
|
|
|
2009-06-17 16:59:54 +00:00
|
|
|
context = _clutter_context_get_default ();
|
2007-11-18 23:24:55 +00:00
|
|
|
|
2007-12-17 17:45:49 +00:00
|
|
|
if (CLUTTER_ACTOR_IS_MAPPED (self) &&
|
|
|
|
(G_UNLIKELY (context->pick_mode == CLUTTER_PICK_ALL) ||
|
|
|
|
CLUTTER_ACTOR_IS_REACTIVE (self)))
|
2007-11-18 23:24:55 +00:00
|
|
|
return TRUE;
|
|
|
|
|
|
|
|
return FALSE;
|
2007-05-28 18:49:34 +00:00
|
|
|
}
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_real_get_preferred_width (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat for_height,
|
|
|
|
gfloat *min_width_p,
|
|
|
|
gfloat *natural_width_p)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
|
|
|
/* Default implementation is always 0x0, usually an actor
|
|
|
|
* using this default is relying on someone to set the
|
|
|
|
* request manually
|
|
|
|
*/
|
2009-04-15 16:25:50 +00:00
|
|
|
CLUTTER_NOTE (LAYOUT, "Default preferred width: 0, 0");
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
if (min_width_p)
|
|
|
|
*min_width_p = 0;
|
|
|
|
|
|
|
|
if (natural_width_p)
|
|
|
|
*natural_width_p = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_real_get_preferred_height (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat for_width,
|
|
|
|
gfloat *min_height_p,
|
|
|
|
gfloat *natural_height_p)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
|
|
|
/* Default implementation is always 0x0, usually an actor
|
|
|
|
* using this default is relying on someone to set the
|
|
|
|
* request manually
|
|
|
|
*/
|
2009-04-15 16:25:50 +00:00
|
|
|
CLUTTER_NOTE (LAYOUT, "Default preferred height: 0, 0");
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
if (min_height_p)
|
|
|
|
*min_height_p = 0;
|
|
|
|
|
|
|
|
if (natural_height_p)
|
|
|
|
*natural_height_p = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_store_old_geometry (ClutterActor *self,
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox *box)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
2008-09-23 11:15:38 +00:00
|
|
|
*box = self->priv->allocation;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline void
|
|
|
|
clutter_actor_notify_if_geometry_changed (ClutterActor *self,
|
2008-06-11 15:08:03 +00:00
|
|
|
const ClutterActorBox *old)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
2009-03-13 14:14:00 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
GObject *obj = G_OBJECT (self);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_freeze_notify (obj);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2009-03-13 14:14:00 +00:00
|
|
|
/* to avoid excessive requisition or allocation cycles we
|
|
|
|
* use the cached values.
|
|
|
|
*
|
|
|
|
* - if we don't have an allocation we assume that we need
|
|
|
|
* to notify anyway
|
|
|
|
* - if we don't have a width or a height request we notify
|
|
|
|
* width and height
|
|
|
|
* - if we have a valid allocation then we check the old
|
|
|
|
* bounding box with the current allocation and we notify
|
|
|
|
* the changes
|
|
|
|
*/
|
|
|
|
if (priv->needs_allocation)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_notify (obj, "x");
|
|
|
|
g_object_notify (obj, "y");
|
|
|
|
g_object_notify (obj, "width");
|
|
|
|
g_object_notify (obj, "height");
|
2009-03-13 14:14:00 +00:00
|
|
|
}
|
|
|
|
else if (priv->needs_width_request || priv->needs_height_request)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_notify (obj, "width");
|
|
|
|
g_object_notify (obj, "height");
|
2009-03-13 14:14:00 +00:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat xu, yu;
|
|
|
|
gfloat widthu, heightu;
|
2009-03-13 14:14:00 +00:00
|
|
|
|
|
|
|
xu = priv->allocation.x1;
|
|
|
|
yu = priv->allocation.y1;
|
|
|
|
widthu = priv->allocation.x2 - priv->allocation.x1;
|
|
|
|
heightu = priv->allocation.y2 - priv->allocation.y1;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2009-03-13 14:14:00 +00:00
|
|
|
if (xu != old->x1)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_notify (obj, "x");
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2009-03-13 14:14:00 +00:00
|
|
|
if (yu != old->y1)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_notify (obj, "y");
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2009-03-13 14:14:00 +00:00
|
|
|
if (widthu != (old->x2 - old->x1))
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_notify (obj, "width");
|
2009-03-13 14:14:00 +00:00
|
|
|
|
|
|
|
if (heightu != (old->y2 - old->y1))
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_notify (obj, "height");
|
2009-03-13 14:14:00 +00:00
|
|
|
}
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_thaw_notify (obj);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
2009-06-03 13:02:06 +00:00
|
|
|
clutter_actor_real_allocate (ClutterActor *self,
|
|
|
|
const ClutterActorBox *box,
|
|
|
|
ClutterAllocationFlags flags)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv = self->priv;
|
|
|
|
gboolean x1_changed, y1_changed, x2_changed, y2_changed;
|
2009-06-04 11:00:58 +00:00
|
|
|
gboolean flags_changed;
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
|
|
|
|
|
|
|
x1_changed = priv->allocation.x1 != box->x1;
|
|
|
|
y1_changed = priv->allocation.y1 != box->y1;
|
|
|
|
x2_changed = priv->allocation.x2 != box->x2;
|
|
|
|
y2_changed = priv->allocation.y2 != box->y2;
|
|
|
|
|
2009-06-04 11:00:58 +00:00
|
|
|
flags_changed = priv->allocation_flags != flags;
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->allocation = *box;
|
2009-06-04 11:00:58 +00:00
|
|
|
priv->allocation_flags = flags;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->needs_allocation = FALSE;
|
|
|
|
|
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
if (x1_changed || y1_changed || x2_changed || y2_changed)
|
2009-06-04 11:00:58 +00:00
|
|
|
{
|
|
|
|
g_object_notify (G_OBJECT (self), "allocation");
|
|
|
|
|
|
|
|
/* we also emit the ::allocation-changed signal for people
|
|
|
|
* that wish to track the allocation flags
|
|
|
|
*/
|
|
|
|
g_signal_emit (self, actor_signals[ALLOCATION_CHANGED], 0,
|
|
|
|
box,
|
|
|
|
flags);
|
|
|
|
}
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
}
|
|
|
|
|
2009-02-17 17:22:02 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_queue_redraw_with_origin (ClutterActor *self,
|
|
|
|
ClutterActor *origin)
|
|
|
|
{
|
|
|
|
/* already queued since last paint() */
|
|
|
|
if (self->priv->queued_redraw)
|
|
|
|
return;
|
|
|
|
|
|
|
|
/* calls klass->queue_redraw in default handler */
|
|
|
|
g_signal_emit (self, actor_signals[QUEUE_REDRAW], 0, origin);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_real_queue_redraw (ClutterActor *self,
|
|
|
|
ClutterActor *origin)
|
|
|
|
{
|
|
|
|
ClutterActor *parent;
|
|
|
|
|
|
|
|
/* already queued since last paint() */
|
|
|
|
if (self->priv->queued_redraw)
|
|
|
|
return;
|
|
|
|
|
2009-10-23 09:08:02 +00:00
|
|
|
/* no point in queuing a paint on a destroyed actor */
|
|
|
|
if (CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_DESTRUCTION)
|
|
|
|
return;
|
|
|
|
|
2009-05-07 18:14:29 +00:00
|
|
|
CLUTTER_NOTE (PAINT, "Redraw queued on '%s'",
|
|
|
|
clutter_actor_get_name (self) ? clutter_actor_get_name (self)
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
2009-02-17 17:22:02 +00:00
|
|
|
self->priv->queued_redraw = TRUE;
|
|
|
|
|
2009-06-11 22:56:27 +00:00
|
|
|
/* If the actor isn't visible, we still had to emit the signal
|
|
|
|
* to allow for a ClutterClone, but the appearance of the parent
|
|
|
|
* won't change so we don't have to propagate up the hierarchy.
|
|
|
|
*/
|
|
|
|
if (!CLUTTER_ACTOR_IS_VISIBLE (self))
|
|
|
|
return;
|
|
|
|
|
2009-02-17 17:22:02 +00:00
|
|
|
/* notify parents, if they are all visible eventually we'll
|
|
|
|
* queue redraw on the stage, which queues the redraw idle.
|
|
|
|
*/
|
|
|
|
parent = clutter_actor_get_parent (self);
|
|
|
|
if (parent != NULL)
|
|
|
|
{
|
|
|
|
/* this will go up recursively */
|
|
|
|
clutter_actor_queue_redraw_with_origin (parent, origin);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-08-08 01:16:04 +00:00
|
|
|
void
|
|
|
|
clutter_actor_real_queue_relayout (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv = self->priv;
|
|
|
|
|
2009-10-23 09:08:02 +00:00
|
|
|
/* no point in queueing a redraw on a destroyed actor */
|
|
|
|
if (CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_DESTRUCTION)
|
|
|
|
return;
|
|
|
|
|
2009-08-08 01:16:04 +00:00
|
|
|
priv->needs_width_request = TRUE;
|
|
|
|
priv->needs_height_request = TRUE;
|
|
|
|
priv->needs_allocation = TRUE;
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
/* reset the cached size requests */
|
|
|
|
memset (priv->width_requests, 0,
|
|
|
|
N_CACHED_SIZE_REQUESTS * sizeof (SizeRequest));
|
|
|
|
memset (priv->height_requests, 0,
|
|
|
|
N_CACHED_SIZE_REQUESTS * sizeof (SizeRequest));
|
|
|
|
|
2009-08-08 01:16:04 +00:00
|
|
|
/* always repaint also (no-op if not mapped) */
|
|
|
|
clutter_actor_queue_redraw (self);
|
|
|
|
|
|
|
|
/* We need to go all the way up the hierarchy */
|
|
|
|
if (priv->parent_actor)
|
|
|
|
clutter_actor_queue_relayout (priv->parent_actor);
|
|
|
|
}
|
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
/* like ClutterVertex, but with a w component */
|
2008-12-19 12:53:57 +00:00
|
|
|
typedef struct {
|
2009-02-23 12:55:01 +00:00
|
|
|
gfloat x;
|
|
|
|
gfloat y;
|
|
|
|
gfloat z;
|
|
|
|
gfloat w;
|
|
|
|
} full_vertex_t;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
/* copies a fixed vertex into a ClutterVertex */
|
|
|
|
static inline void
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_to_units (const full_vertex_t *f,
|
|
|
|
ClutterVertex *u)
|
2008-12-19 12:53:57 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
u->x = f->x;
|
|
|
|
u->y = f->y;
|
|
|
|
u->z = f->z;
|
2008-12-19 12:53:57 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Transforms a vertex using the passed matrix; vertex is
|
|
|
|
* an in-out parameter
|
|
|
|
*/
|
2009-02-18 18:54:54 +00:00
|
|
|
static void
|
|
|
|
mtx_transform (const CoglMatrix *matrix,
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_t *vertex)
|
2008-12-19 12:53:57 +00:00
|
|
|
{
|
2009-02-18 18:54:54 +00:00
|
|
|
cogl_matrix_transform_point (matrix,
|
|
|
|
&vertex->x,
|
|
|
|
&vertex->y,
|
|
|
|
&vertex->z,
|
|
|
|
&vertex->w);
|
2007-05-31 17:11:09 +00:00
|
|
|
}
|
|
|
|
|
2008-12-19 12:53:57 +00:00
|
|
|
/* Help macros to scale from OpenGL <-1,1> coordinates system to our
|
|
|
|
* X-window based <0,window-size> coordinates
|
|
|
|
*/
|
2009-02-23 12:55:01 +00:00
|
|
|
#define MTX_GL_SCALE_X(x,w,v1,v2) ((((((x) / (w)) + 1.0) / 2) * (v1)) + (v2))
|
|
|
|
#define MTX_GL_SCALE_Y(y,w,v1,v2) ((v1) - (((((y) / (w)) + 1.0) / 2) * (v1)) + (v2))
|
|
|
|
#define MTX_GL_SCALE_Z(z,w,v1,v2) (MTX_GL_SCALE_X ((z), (w), (v1), (v2)))
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
/* transforms a 4-tuple of coordinates using @matrix and
|
|
|
|
* places the result into a fixed @vertex
|
|
|
|
*/
|
|
|
|
static inline void
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_transform (const CoglMatrix *matrix,
|
|
|
|
gfloat x,
|
|
|
|
gfloat y,
|
|
|
|
gfloat z,
|
|
|
|
gfloat w,
|
|
|
|
full_vertex_t *vertex)
|
2008-12-19 12:53:57 +00:00
|
|
|
{
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_t tmp = { 0, };
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
tmp.x = x;
|
|
|
|
tmp.y = y;
|
|
|
|
tmp.z = z;
|
|
|
|
tmp.w = w;
|
|
|
|
|
|
|
|
mtx_transform (matrix, &tmp);
|
|
|
|
|
|
|
|
*vertex = tmp;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* scales a fixed @vertex using @matrix and @viewport, and
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* transforms the result into a ClutterVertex, filling @vertex_p
|
2008-12-19 12:53:57 +00:00
|
|
|
*/
|
|
|
|
static inline void
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_scale (const CoglMatrix *matrix,
|
|
|
|
const full_vertex_t *vertex,
|
|
|
|
const gfloat viewport[],
|
|
|
|
ClutterVertex *vertex_p)
|
2008-12-19 12:53:57 +00:00
|
|
|
{
|
2009-02-23 12:55:01 +00:00
|
|
|
gfloat v_x, v_y, v_width, v_height;
|
|
|
|
full_vertex_t tmp = { 0, };
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
tmp = *vertex;
|
|
|
|
|
|
|
|
mtx_transform (matrix, &tmp);
|
|
|
|
|
|
|
|
v_x = viewport[0];
|
|
|
|
v_y = viewport[1];
|
|
|
|
v_width = viewport[2];
|
|
|
|
v_height = viewport[3];
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
tmp.x = MTX_GL_SCALE_X (tmp.x, tmp.w, v_width, v_x);
|
2008-12-19 12:53:57 +00:00
|
|
|
tmp.y = MTX_GL_SCALE_Y (tmp.y, tmp.w, v_height, v_y);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
tmp.z = MTX_GL_SCALE_Z (tmp.z, tmp.w, v_width, v_x);
|
2008-12-19 12:53:57 +00:00
|
|
|
tmp.w = 0;
|
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_to_units (&tmp, vertex_p);
|
2008-12-19 12:53:57 +00:00
|
|
|
}
|
|
|
|
|
2008-02-14 15:59:38 +00:00
|
|
|
/* Applies the transforms associated with this actor and its ancestors,
|
|
|
|
* retrieves the resulting OpenGL modelview matrix, and uses the matrix
|
|
|
|
* to transform the supplied point
|
2008-12-19 12:53:57 +00:00
|
|
|
*
|
|
|
|
* The point coordinates are in-out parameters
|
2008-02-14 15:59:38 +00:00
|
|
|
*/
|
|
|
|
static void
|
|
|
|
clutter_actor_transform_point_relative (ClutterActor *actor,
|
|
|
|
ClutterActor *ancestor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat *x,
|
|
|
|
gfloat *y,
|
|
|
|
gfloat *z,
|
|
|
|
gfloat *w)
|
2008-02-14 15:59:38 +00:00
|
|
|
{
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_t vertex = { 0, };
|
2009-02-18 18:54:54 +00:00
|
|
|
CoglMatrix matrix;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
vertex.x = (x != NULL) ? *x : 0;
|
|
|
|
vertex.y = (y != NULL) ? *y : 0;
|
|
|
|
vertex.z = (z != NULL) ? *z : 0;
|
|
|
|
vertex.w = (w != NULL) ? *w : 0;
|
2008-02-14 15:59:38 +00:00
|
|
|
|
|
|
|
cogl_push_matrix();
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
_clutter_actor_apply_modelview_transform_recursive (actor, ancestor);
|
2008-12-19 12:53:57 +00:00
|
|
|
|
2009-02-18 18:54:54 +00:00
|
|
|
cogl_get_modelview_matrix (&matrix);
|
|
|
|
mtx_transform (&matrix, &vertex);
|
2008-02-14 15:59:38 +00:00
|
|
|
|
|
|
|
cogl_pop_matrix();
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
if (x)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*x = vertex.x;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
if (y)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*y = vertex.y;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
if (z)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*z = vertex.z;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
if (w)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*w = vertex.w;
|
2008-02-14 15:59:38 +00:00
|
|
|
}
|
|
|
|
|
2007-06-15 09:53:07 +00:00
|
|
|
/* Applies the transforms associated with this actor and its ancestors,
|
|
|
|
* retrieves the resulting OpenGL modelview matrix, and uses the matrix
|
|
|
|
* to transform the supplied point
|
|
|
|
*/
|
2007-06-01 16:20:57 +00:00
|
|
|
static void
|
2007-10-12 08:17:00 +00:00
|
|
|
clutter_actor_transform_point (ClutterActor *actor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat *x,
|
|
|
|
gfloat *y,
|
|
|
|
gfloat *z,
|
|
|
|
gfloat *w)
|
2007-06-01 16:20:57 +00:00
|
|
|
{
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_t vertex = { 0, };
|
2009-02-18 18:54:54 +00:00
|
|
|
CoglMatrix matrix;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
vertex.x = (x != NULL) ? *x : 0;
|
|
|
|
vertex.y = (y != NULL) ? *y : 0;
|
|
|
|
vertex.z = (z != NULL) ? *z : 0;
|
|
|
|
vertex.w = (w != NULL) ? *w : 0;
|
2007-06-01 16:20:57 +00:00
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
cogl_push_matrix();
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
_clutter_actor_apply_modelview_transform_recursive (actor, NULL);
|
2008-12-19 12:53:57 +00:00
|
|
|
|
2009-02-18 18:54:54 +00:00
|
|
|
cogl_get_modelview_matrix (&matrix);
|
|
|
|
mtx_transform (&matrix, &vertex);
|
2007-06-12 11:42:29 +00:00
|
|
|
|
|
|
|
cogl_pop_matrix();
|
2007-06-01 16:20:57 +00:00
|
|
|
|
2008-12-19 12:53:57 +00:00
|
|
|
if (x)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*x = vertex.x;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
if (y)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*y = vertex.y;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
if (z)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*z = vertex.z;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
|
|
|
if (w)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*w = vertex.w;
|
2008-12-19 12:53:57 +00:00
|
|
|
}
|
2007-06-12 11:42:29 +00:00
|
|
|
|
2008-02-14 15:59:38 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_apply_relative_transform_to_point:
|
|
|
|
* @self: A #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @ancestor: (allow-none): A #ClutterActor ancestor, or %NULL to use the
|
2008-02-18 09:43:01 +00:00
|
|
|
* default #ClutterStage
|
2008-02-14 15:59:38 +00:00
|
|
|
* @point: A point as #ClutterVertex
|
|
|
|
* @vertex: The translated #ClutterVertex
|
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* Transforms @point in coordinates relative to the actor into
|
|
|
|
* ancestor-relative coordinates using the relevant transform
|
2008-02-18 09:43:01 +00:00
|
|
|
* stack (i.e. scale, rotation, etc).
|
|
|
|
*
|
|
|
|
* If @ancestor is %NULL the ancestor will be the #ClutterStage. In
|
|
|
|
* this case, the coordinates returned will be the coordinates on
|
|
|
|
* the stage before the projection is applied. This is different from
|
|
|
|
* the behaviour of clutter_actor_apply_transform_to_point().
|
2008-02-14 15:59:38 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
2008-02-18 09:43:01 +00:00
|
|
|
*/
|
2008-02-14 15:59:38 +00:00
|
|
|
void
|
2008-12-10 23:00:48 +00:00
|
|
|
clutter_actor_apply_relative_transform_to_point (ClutterActor *self,
|
|
|
|
ClutterActor *ancestor,
|
|
|
|
const ClutterVertex *point,
|
|
|
|
ClutterVertex *vertex)
|
2008-02-14 15:59:38 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat x, y, z, w;
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_t tmp;
|
|
|
|
gfloat v[4];
|
2008-02-14 15:59:38 +00:00
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2008-02-18 09:43:01 +00:00
|
|
|
g_return_if_fail (ancestor == NULL || CLUTTER_IS_ACTOR (ancestor));
|
2008-12-10 23:00:48 +00:00
|
|
|
g_return_if_fail (point != NULL);
|
|
|
|
g_return_if_fail (vertex != NULL);
|
|
|
|
|
2009-08-02 17:34:17 +00:00
|
|
|
x = point->x;
|
|
|
|
y = point->y;
|
|
|
|
z = point->z;
|
2009-01-20 16:20:54 +00:00
|
|
|
w = 1.0;
|
2008-02-14 15:59:38 +00:00
|
|
|
|
|
|
|
/* First we tranform the point using the OpenGL modelview matrix */
|
2009-01-08 15:45:22 +00:00
|
|
|
clutter_actor_transform_point_relative (self, ancestor, &x, &y, &z, &w);
|
2008-02-14 15:59:38 +00:00
|
|
|
|
|
|
|
cogl_get_viewport (v);
|
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
/* The w[3] parameter should always be 1.0 here, so we ignore it; otherwise
|
2008-02-14 15:59:38 +00:00
|
|
|
* we would have to divide the original verts with it.
|
|
|
|
*/
|
2009-02-23 12:55:01 +00:00
|
|
|
tmp.x = (x + 0.5) * v[2];
|
|
|
|
tmp.y = (0.5 - y) * v[3];
|
|
|
|
tmp.z = (z + 0.5) * v[2];
|
2009-01-08 15:45:22 +00:00
|
|
|
tmp.w = 0;
|
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_to_units (&tmp, vertex);
|
2008-02-14 15:59:38 +00:00
|
|
|
}
|
|
|
|
|
2007-05-31 17:11:09 +00:00
|
|
|
/**
|
2007-06-27 15:10:52 +00:00
|
|
|
* clutter_actor_apply_transform_to_point:
|
2007-05-31 17:11:09 +00:00
|
|
|
* @self: A #ClutterActor
|
2007-07-02 09:21:58 +00:00
|
|
|
* @point: A point as #ClutterVertex
|
|
|
|
* @vertex: The translated #ClutterVertex
|
2007-05-31 17:11:09 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* Transforms @point in coordinates relative to the actor
|
|
|
|
* into screen-relative coordinates with the current actor
|
2008-02-15 15:18:02 +00:00
|
|
|
* transformation (i.e. scale, rotation, etc)
|
2007-05-31 17:11:09 +00:00
|
|
|
*
|
|
|
|
* Since: 0.4
|
|
|
|
**/
|
|
|
|
void
|
2008-12-10 23:00:48 +00:00
|
|
|
clutter_actor_apply_transform_to_point (ClutterActor *self,
|
|
|
|
const ClutterVertex *point,
|
|
|
|
ClutterVertex *vertex)
|
2007-06-12 11:42:29 +00:00
|
|
|
{
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_t tmp = { 0, };
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat x, y, z, w;
|
2009-02-18 18:54:54 +00:00
|
|
|
CoglMatrix matrix_p;
|
2009-02-23 12:55:01 +00:00
|
|
|
gfloat v[4];
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2008-12-10 23:00:48 +00:00
|
|
|
g_return_if_fail (point != NULL);
|
|
|
|
g_return_if_fail (vertex != NULL);
|
|
|
|
|
2009-01-08 15:45:22 +00:00
|
|
|
x = point->x;
|
|
|
|
y = point->y;
|
|
|
|
z = point->z;
|
2009-02-23 12:55:01 +00:00
|
|
|
w = 1.0;
|
2007-05-31 17:11:09 +00:00
|
|
|
|
2007-06-15 09:53:07 +00:00
|
|
|
/* First we tranform the point using the OpenGL modelview matrix */
|
2009-01-08 15:45:22 +00:00
|
|
|
clutter_actor_transform_point (self, &x, &y, &z, &w);
|
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
tmp.x = x;
|
|
|
|
tmp.y = y;
|
|
|
|
tmp.z = z;
|
|
|
|
tmp.w = w;
|
2007-05-31 18:22:41 +00:00
|
|
|
|
2009-02-18 18:54:54 +00:00
|
|
|
cogl_get_projection_matrix (&matrix_p);
|
2007-06-12 11:42:29 +00:00
|
|
|
cogl_get_viewport (v);
|
2007-06-15 09:53:07 +00:00
|
|
|
|
|
|
|
/* Now, transform it again with the projection matrix */
|
2009-02-18 18:54:54 +00:00
|
|
|
mtx_transform (&matrix_p, &tmp);
|
2007-05-31 17:11:09 +00:00
|
|
|
|
2007-06-15 09:53:07 +00:00
|
|
|
/* Finaly translate from OpenGL coords to window coords */
|
2009-02-23 12:55:01 +00:00
|
|
|
vertex->x = MTX_GL_SCALE_X (tmp.x, tmp.w, v[2], v[0]);
|
|
|
|
vertex->y = MTX_GL_SCALE_Y (tmp.y, tmp.w, v[3], v[1]);
|
|
|
|
vertex->z = MTX_GL_SCALE_Z (tmp.z, tmp.w, v[2], v[0]);
|
2007-05-31 17:11:09 +00:00
|
|
|
}
|
|
|
|
|
2008-02-14 15:59:38 +00:00
|
|
|
/* Recursively tranform supplied vertices with the tranform for the current
|
|
|
|
* actor and up to the ancestor (like clutter_actor_transform_point() but
|
|
|
|
* for all the vertices in one go).
|
|
|
|
*/
|
|
|
|
static void
|
2009-02-23 12:55:01 +00:00
|
|
|
clutter_actor_transform_vertices_relative (ClutterActor *self,
|
|
|
|
ClutterActor *ancestor,
|
|
|
|
full_vertex_t vertices[])
|
2008-02-14 15:59:38 +00:00
|
|
|
{
|
2008-12-19 12:53:57 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2009-02-23 12:55:01 +00:00
|
|
|
gfloat width, height;
|
2009-02-18 18:54:54 +00:00
|
|
|
CoglMatrix mtx;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
width = priv->allocation.x2 - priv->allocation.x1;
|
|
|
|
height = priv->allocation.y2 - priv->allocation.y1;
|
2008-02-14 15:59:38 +00:00
|
|
|
|
|
|
|
cogl_push_matrix();
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
_clutter_actor_apply_modelview_transform_recursive (self, ancestor);
|
2008-02-14 15:59:38 +00:00
|
|
|
|
2009-02-18 18:54:54 +00:00
|
|
|
cogl_get_modelview_matrix (&mtx);
|
2008-02-14 15:59:38 +00:00
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_transform (&mtx, 0, 0, 0, 1.0, &vertices[0]);
|
|
|
|
full_vertex_transform (&mtx, width, 0, 0, 1.0, &vertices[1]);
|
|
|
|
full_vertex_transform (&mtx, 0, height, 0, 1.0, &vertices[2]);
|
|
|
|
full_vertex_transform (&mtx, width, height, 0, 1.0, &vertices[3]);
|
2008-02-14 15:59:38 +00:00
|
|
|
|
|
|
|
cogl_pop_matrix();
|
|
|
|
}
|
|
|
|
|
2008-07-01 21:52:19 +00:00
|
|
|
/* Recursively tranform supplied box with the tranform for the current
|
|
|
|
* actor and all its ancestors (like clutter_actor_transform_point()
|
|
|
|
* but for all the vertices in one go) and project it into screen
|
|
|
|
* coordinates
|
2007-06-12 11:42:29 +00:00
|
|
|
*/
|
|
|
|
static void
|
2008-07-01 21:52:19 +00:00
|
|
|
clutter_actor_transform_and_project_box (ClutterActor *self,
|
|
|
|
const ClutterActorBox *box,
|
2009-07-27 15:10:07 +00:00
|
|
|
ClutterVertex verts[])
|
2007-05-31 17:11:09 +00:00
|
|
|
{
|
2008-12-19 12:53:57 +00:00
|
|
|
ClutterActor *stage;
|
2009-02-18 18:54:54 +00:00
|
|
|
CoglMatrix mtx;
|
|
|
|
CoglMatrix mtx_p;
|
2009-02-23 12:55:01 +00:00
|
|
|
gfloat v[4];
|
|
|
|
gfloat width, height;
|
|
|
|
full_vertex_t vertices[4];
|
2008-12-19 12:53:57 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
width = box->x2 - box->x1;
|
|
|
|
height = box->y2 - box->y1;
|
2008-07-01 21:52:19 +00:00
|
|
|
|
|
|
|
/* We essentially have to dupe some code from clutter_redraw() here
|
|
|
|
* to make sure GL Matrices etc are initialised if we're called and we
|
|
|
|
* havn't yet rendered anything.
|
|
|
|
*
|
|
|
|
* Simply duping code for now in wait for Cogl cleanup that can hopefully
|
|
|
|
* address this in a nicer way.
|
|
|
|
*/
|
|
|
|
stage = clutter_actor_get_stage (self);
|
|
|
|
|
|
|
|
/* FIXME: if were not yet added to a stage, its probably unsafe to
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* return default - ideally the func should fail
|
2008-07-01 21:52:19 +00:00
|
|
|
*/
|
|
|
|
if (stage == NULL)
|
|
|
|
stage = clutter_stage_get_default ();
|
|
|
|
|
|
|
|
clutter_stage_ensure_current (CLUTTER_STAGE (stage));
|
|
|
|
_clutter_stage_maybe_setup_viewport (CLUTTER_STAGE (stage));
|
2007-05-31 17:11:09 +00:00
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
cogl_push_matrix();
|
2008-12-19 12:53:57 +00:00
|
|
|
|
2008-02-14 15:59:38 +00:00
|
|
|
_clutter_actor_apply_modelview_transform_recursive (self, NULL);
|
2007-06-01 16:20:57 +00:00
|
|
|
|
2009-02-18 18:54:54 +00:00
|
|
|
cogl_get_modelview_matrix (&mtx);
|
2007-05-31 09:29:12 +00:00
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_transform (&mtx, 0, 0, 0, 1.0, &vertices[0]);
|
|
|
|
full_vertex_transform (&mtx, width, 0, 0, 1.0, &vertices[1]);
|
|
|
|
full_vertex_transform (&mtx, 0, height, 0, 1.0, &vertices[2]);
|
|
|
|
full_vertex_transform (&mtx, width, height, 0, 1.0, &vertices[3]);
|
2007-05-31 09:29:12 +00:00
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
cogl_pop_matrix();
|
2008-07-01 21:52:19 +00:00
|
|
|
|
2009-02-18 18:54:54 +00:00
|
|
|
cogl_get_projection_matrix (&mtx_p);
|
2008-07-01 21:52:19 +00:00
|
|
|
cogl_get_viewport (v);
|
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
full_vertex_scale (&mtx_p, &vertices[0], v, &verts[0]);
|
|
|
|
full_vertex_scale (&mtx_p, &vertices[1], v, &verts[1]);
|
|
|
|
full_vertex_scale (&mtx_p, &vertices[2], v, &verts[2]);
|
|
|
|
full_vertex_scale (&mtx_p, &vertices[3], v, &verts[3]);
|
2007-05-31 09:29:12 +00:00
|
|
|
}
|
|
|
|
|
2008-02-14 15:59:38 +00:00
|
|
|
/**
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* clutter_actor_get_allocation_vertices:
|
2008-02-14 15:59:38 +00:00
|
|
|
* @self: A #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @ancestor: (allow-none): A #ClutterActor to calculate the vertices
|
|
|
|
* against, or %NULL to use the default #ClutterStage
|
2009-06-10 13:54:42 +00:00
|
|
|
* @verts: (out) (array fixed-size=4): return location for an array of
|
|
|
|
* 4 #ClutterVertex in which to store the result.
|
2008-02-14 15:59:38 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* Calculates the transformed coordinates of the four corners of the
|
2008-02-18 09:43:01 +00:00
|
|
|
* actor in the plane of @ancestor. The returned vertices relate to
|
2008-02-15 14:39:25 +00:00
|
|
|
* the #ClutterActorBox coordinates as follows:
|
|
|
|
* <itemizedlist>
|
2008-12-19 12:53:57 +00:00
|
|
|
* <listitem><para>@verts[0] contains (x1, y1)</para></listitem>
|
|
|
|
* <listitem><para>@verts[1] contains (x2, y1)</para></listitem>
|
|
|
|
* <listitem><para>@verts[2] contains (x1, y2)</para></listitem>
|
|
|
|
* <listitem><para>@verts[3] contains (x2, y2)</para></listitem>
|
2008-02-15 14:39:25 +00:00
|
|
|
* </itemizedlist>
|
2008-02-14 15:59:38 +00:00
|
|
|
*
|
2008-02-18 09:43:01 +00:00
|
|
|
* If @ancestor is %NULL the ancestor will be the #ClutterStage. In
|
|
|
|
* this case, the coordinates returned will be the coordinates on
|
|
|
|
* the stage before the projection is applied. This is different from
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* the behaviour of clutter_actor_get_abs_allocation_vertices().
|
2008-02-18 09:43:01 +00:00
|
|
|
*
|
2008-02-14 15:59:38 +00:00
|
|
|
* Since: 0.6
|
2008-02-15 14:39:25 +00:00
|
|
|
*/
|
2008-02-14 15:59:38 +00:00
|
|
|
void
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_get_allocation_vertices (ClutterActor *self,
|
|
|
|
ClutterActor *ancestor,
|
2009-07-27 15:10:07 +00:00
|
|
|
ClutterVertex verts[])
|
2008-02-14 15:59:38 +00:00
|
|
|
{
|
2008-12-19 12:53:57 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
ClutterActor *stage;
|
2009-02-23 12:55:01 +00:00
|
|
|
gfloat v[4];
|
|
|
|
full_vertex_t vertices[4];
|
|
|
|
full_vertex_t tmp = { 0, };
|
2008-02-14 15:59:38 +00:00
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2008-02-18 09:43:01 +00:00
|
|
|
g_return_if_fail (ancestor == NULL || CLUTTER_IS_ACTOR (ancestor));
|
2008-02-14 15:59:38 +00:00
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
/* We essentially have to dupe some code from clutter_redraw() here
|
|
|
|
* to make sure GL Matrices etc are initialised if we're called and we
|
|
|
|
* havn't yet rendered anything.
|
|
|
|
*
|
|
|
|
* Simply duping code for now in wait for Cogl cleanup that can hopefully
|
|
|
|
* address this in a nicer way.
|
2009-02-23 12:55:01 +00:00
|
|
|
*/
|
2008-03-28 22:50:55 +00:00
|
|
|
stage = clutter_actor_get_stage (self);
|
|
|
|
|
|
|
|
/* FIXME: if were not yet added to a stage, its probably unsafe to
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* return default - idealy the func should fail
|
2008-03-28 22:50:55 +00:00
|
|
|
*/
|
|
|
|
if (stage == NULL)
|
|
|
|
stage = clutter_stage_get_default ();
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_stage_ensure_current (CLUTTER_STAGE (stage));
|
|
|
|
_clutter_stage_maybe_setup_viewport (CLUTTER_STAGE (stage));
|
2008-02-14 15:59:38 +00:00
|
|
|
|
2008-07-01 21:52:19 +00:00
|
|
|
/* if the actor needs to be allocated we force a relayout, so that
|
|
|
|
* clutter_actor_transform_vertices_relative() will have valid values
|
|
|
|
* to use in the transformations
|
|
|
|
*/
|
|
|
|
if (priv->needs_allocation)
|
|
|
|
_clutter_stage_maybe_relayout (stage);
|
|
|
|
|
2008-12-19 12:53:57 +00:00
|
|
|
clutter_actor_transform_vertices_relative (self, ancestor, vertices);
|
|
|
|
|
2008-02-14 15:59:38 +00:00
|
|
|
cogl_get_viewport (v);
|
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
/* The w[3] parameter should always be 1.0 here, so we ignore it;
|
2008-12-19 12:53:57 +00:00
|
|
|
* otherwise we would have to divide the original verts with it.
|
2008-02-14 15:59:38 +00:00
|
|
|
*/
|
2009-02-23 12:55:01 +00:00
|
|
|
tmp.x = ((vertices[0].x + 0.5) * v[2]);
|
|
|
|
tmp.y = ((0.5 - vertices[0].y) * v[3]);
|
|
|
|
tmp.z = ((vertices[0].z + 0.5) * v[2]);
|
|
|
|
full_vertex_to_units (&tmp, &verts[0]);
|
2008-02-14 15:59:38 +00:00
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
tmp.x = ((vertices[1].x + 0.5) * v[2]);
|
|
|
|
tmp.y = ((0.5 - vertices[1].y) * v[3]);
|
|
|
|
tmp.z = ((vertices[1].z + 0.5) * v[2]);
|
|
|
|
full_vertex_to_units (&tmp, &verts[1]);
|
2008-02-14 15:59:38 +00:00
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
tmp.x = ((vertices[2].x + 0.5) * v[2]);
|
|
|
|
tmp.y = ((0.5 - vertices[2].y) * v[3]);
|
|
|
|
tmp.z = ((vertices[2].z + 0.5) * v[2]);
|
|
|
|
full_vertex_to_units (&tmp, &verts[2]);
|
2008-02-14 15:59:38 +00:00
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
tmp.x = ((vertices[3].x + 0.5) * v[2]);
|
|
|
|
tmp.y = ((0.5 - vertices[3].y) * v[3]);
|
|
|
|
tmp.z = ((vertices[3].z + 0.5) * v[2]);
|
|
|
|
full_vertex_to_units (&tmp, &verts[3]);
|
2008-02-14 15:59:38 +00:00
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* clutter_actor_get_abs_allocation_vertices:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
2009-06-10 13:54:42 +00:00
|
|
|
* @verts: (out) (array fixed-size=4): Pointer to a location of an array
|
|
|
|
* of 4 #ClutterVertex where to store the result.
|
2007-10-12 08:17:00 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* Calculates the transformed screen coordinates of the four corners of
|
2008-02-18 09:43:01 +00:00
|
|
|
* the actor; the returned vertices relate to the #ClutterActorBox
|
2007-11-21 23:25:30 +00:00
|
|
|
* coordinates as follows:
|
2008-02-18 09:43:01 +00:00
|
|
|
* <itemizedlist>
|
|
|
|
* <listitem><para>v[0] contains (x1, y1)</para></listitem>
|
|
|
|
* <listitem><para>v[1] contains (x2, y1)</para></listitem>
|
|
|
|
* <listitem><para>v[2] contains (x1, y2)</para></listitem>
|
|
|
|
* <listitem><para>v[3] contains (x2, y2)</para></listitem>
|
|
|
|
* </itemizedlist>
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2007-06-12 11:42:29 +00:00
|
|
|
* Since: 0.4
|
2008-02-18 09:43:01 +00:00
|
|
|
*/
|
2006-06-13 13:17:45 +00:00
|
|
|
void
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_get_abs_allocation_vertices (ClutterActor *self,
|
2009-07-27 15:10:07 +00:00
|
|
|
ClutterVertex verts[])
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2007-06-12 11:42:29 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-03-22 18:21:59 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2007-06-12 11:42:29 +00:00
|
|
|
|
2007-03-22 18:21:59 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
2008-07-01 21:52:19 +00:00
|
|
|
/* if the actor needs to be allocated we force a relayout, so that
|
|
|
|
* the actor allocation box will be valid for
|
|
|
|
* clutter_actor_transform_and_project_box()
|
|
|
|
*/
|
|
|
|
if (priv->needs_allocation)
|
2008-07-02 16:15:30 +00:00
|
|
|
{
|
|
|
|
ClutterActor *stage = clutter_actor_get_stage (self);
|
|
|
|
|
|
|
|
/* FIXME: if were not yet added to a stage, its probably unsafe to
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* return default - idealy the func should fail
|
2008-07-02 16:15:30 +00:00
|
|
|
*/
|
|
|
|
if (stage == NULL)
|
|
|
|
stage = clutter_stage_get_default ();
|
|
|
|
|
|
|
|
_clutter_stage_maybe_relayout (stage);
|
|
|
|
}
|
2007-06-12 11:42:29 +00:00
|
|
|
|
2008-07-01 21:52:19 +00:00
|
|
|
clutter_actor_transform_and_project_box (self,
|
|
|
|
&self->priv->allocation,
|
|
|
|
verts);
|
2007-06-12 11:42:29 +00:00
|
|
|
}
|
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
static void
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
clutter_actor_real_apply_transform (ClutterActor *self,
|
|
|
|
CoglMatrix *matrix)
|
2007-06-12 11:42:29 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv = self->priv;
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
gboolean is_stage = CLUTTER_IS_STAGE (self);
|
2007-11-28 12:23:31 +00:00
|
|
|
|
2008-02-07 13:05:18 +00:00
|
|
|
if (!is_stage)
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
{
|
|
|
|
cogl_matrix_translate (matrix,
|
|
|
|
priv->allocation.x1,
|
|
|
|
priv->allocation.y1,
|
|
|
|
0.0);
|
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
if (priv->z)
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
cogl_matrix_translate (matrix, 0, 0, priv->z);
|
2009-01-23 15:55:41 +00:00
|
|
|
|
2007-06-27 11:34:43 +00:00
|
|
|
/*
|
|
|
|
* because the rotation involves translations, we must scale before
|
|
|
|
* applying the rotations (if we apply the scale after the rotations,
|
|
|
|
* the translations included in the rotation are not scaled and so the
|
|
|
|
* entire object will move on the screen as a result of rotating it).
|
|
|
|
*/
|
2009-01-20 16:20:54 +00:00
|
|
|
if (priv->scale_x != 1.0 || priv->scale_y != 1.0)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
{
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
TRANSFORM_ABOUT_ANCHOR_COORD (self, matrix,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&priv->scale_center,
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
cogl_matrix_scale (matrix,
|
|
|
|
priv->scale_x,
|
|
|
|
priv->scale_y,
|
|
|
|
1.0));
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
}
|
2007-11-29 13:47:37 +00:00
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
if (priv->rzang)
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
TRANSFORM_ABOUT_ANCHOR_COORD (self, matrix,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&priv->rz_center,
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
cogl_matrix_rotate (matrix,
|
|
|
|
priv->rzang,
|
|
|
|
0, 0, 1.0));
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
if (priv->ryang)
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
TRANSFORM_ABOUT_ANCHOR_COORD (self, matrix,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&priv->ry_center,
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
cogl_matrix_rotate (matrix,
|
|
|
|
priv->ryang,
|
|
|
|
0, 1.0, 0));
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
if (priv->rxang)
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
TRANSFORM_ABOUT_ANCHOR_COORD (self, matrix,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&priv->rx_center,
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
cogl_matrix_rotate (matrix,
|
|
|
|
priv->rxang,
|
|
|
|
1.0, 0, 0));
|
2007-11-29 13:47:37 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
if (!is_stage && !clutter_anchor_coord_is_zero (&priv->anchor))
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat x, y, z;
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_anchor_coord_get_units (self, &priv->anchor, &x, &y, &z);
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
cogl_matrix_translate (matrix, -x, -y, -z);
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
2007-06-12 11:42:29 +00:00
|
|
|
}
|
|
|
|
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
/* Applies the transforms associated with this actor to the
|
|
|
|
* OpenGL modelview matrix.
|
|
|
|
*
|
|
|
|
* This function does not push/pop matrix; it is the responsibility
|
|
|
|
* of the caller to do so as appropriate
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
_clutter_actor_apply_modelview_transform (ClutterActor *self)
|
|
|
|
{
|
|
|
|
CoglMatrix matrix, cur, new;
|
|
|
|
|
|
|
|
cogl_matrix_init_identity (&matrix);
|
|
|
|
|
|
|
|
clutter_actor_get_transformation_matrix (self, &matrix);
|
|
|
|
|
|
|
|
cogl_get_modelview_matrix (&cur);
|
|
|
|
|
|
|
|
cogl_matrix_multiply (&new, &cur, &matrix);
|
|
|
|
|
|
|
|
cogl_set_modelview_matrix (&new);
|
|
|
|
}
|
|
|
|
|
2007-06-15 09:53:07 +00:00
|
|
|
/* Recursively applies the transforms associated with this actor and
|
2008-02-14 15:59:38 +00:00
|
|
|
* its ancestors to the OpenGL modelview matrix. Use NULL if you want this
|
|
|
|
* to go all the way down to the stage.
|
2007-06-15 09:53:07 +00:00
|
|
|
*
|
|
|
|
* This function does not push/pop matrix; it is the responsibility
|
|
|
|
* of the caller to do so as appropriate
|
|
|
|
*/
|
2008-08-01 12:23:57 +00:00
|
|
|
void
|
2008-02-18 09:43:01 +00:00
|
|
|
_clutter_actor_apply_modelview_transform_recursive (ClutterActor *self,
|
|
|
|
ClutterActor *ancestor)
|
2007-06-12 11:42:29 +00:00
|
|
|
{
|
2008-03-28 22:50:55 +00:00
|
|
|
ClutterActor *parent, *stage;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
parent = clutter_actor_get_parent (self);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-02-14 15:59:38 +00:00
|
|
|
/*
|
|
|
|
* If we reached the ancestor, quit
|
|
|
|
* NB: NULL ancestor means the stage, and this will not trigger
|
|
|
|
* (as it should not)
|
|
|
|
*/
|
|
|
|
if (self == ancestor)
|
|
|
|
return;
|
|
|
|
|
2008-03-28 22:50:55 +00:00
|
|
|
stage = clutter_actor_get_stage (self);
|
|
|
|
|
|
|
|
/* FIXME: if were not yet added to a stage, its probably unsafe to
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* return default - idealy the func should fail
|
2008-03-28 22:50:55 +00:00
|
|
|
*/
|
|
|
|
if (stage == NULL)
|
|
|
|
stage = clutter_stage_get_default ();
|
|
|
|
|
2009-10-23 09:08:02 +00:00
|
|
|
if (parent != NULL)
|
2008-02-14 15:59:38 +00:00
|
|
|
_clutter_actor_apply_modelview_transform_recursive (parent, ancestor);
|
2008-03-28 22:50:55 +00:00
|
|
|
else if (self != stage)
|
|
|
|
_clutter_actor_apply_modelview_transform (stage);
|
2007-07-30 07:27:14 +00:00
|
|
|
|
|
|
|
_clutter_actor_apply_modelview_transform (self);
|
2007-06-12 11:42:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_paint:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
|
|
|
* Renders the actor to display.
|
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* This function should not be called directly by applications.
|
|
|
|
* Call clutter_actor_queue_redraw() to queue paints, instead.
|
2009-01-23 11:34:13 +00:00
|
|
|
*
|
|
|
|
* This function will emit the #ClutterActor::paint signal.
|
2008-02-15 14:39:25 +00:00
|
|
|
*/
|
2007-06-12 11:42:29 +00:00
|
|
|
void
|
|
|
|
clutter_actor_paint (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
ClutterMainContext *context;
|
2008-06-25 08:40:25 +00:00
|
|
|
gboolean clip_set = FALSE;
|
2009-03-30 15:41:02 +00:00
|
|
|
CLUTTER_STATIC_COUNTER (actor_paint_counter,
|
|
|
|
"Actor real-paint counter",
|
|
|
|
"Increments each time any actor is painted",
|
|
|
|
0 /* no application private data */);
|
|
|
|
CLUTTER_STATIC_COUNTER (actor_pick_counter,
|
|
|
|
"Actor pick-paint counter",
|
|
|
|
"Increments each time any actor is painted "
|
|
|
|
"for picking",
|
|
|
|
0 /* no application private data */);
|
2008-06-05 14:32:18 +00:00
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
priv = self->priv;
|
2009-06-17 16:59:54 +00:00
|
|
|
|
|
|
|
context = _clutter_context_get_default ();
|
2009-05-20 16:55:29 +00:00
|
|
|
|
|
|
|
/* It's an important optimization that we consider painting of
|
|
|
|
* actors with 0 opacity to be a NOP... */
|
|
|
|
if (G_LIKELY (context->pick_mode == CLUTTER_PICK_NONE) &&
|
2009-06-05 15:25:39 +00:00
|
|
|
/* If the actor is being painted from a clone then check the
|
|
|
|
clone's opacity instead */
|
|
|
|
(priv->opacity_parent ? priv->opacity_parent->priv : priv)->opacity == 0)
|
2009-05-22 13:49:37 +00:00
|
|
|
{
|
|
|
|
priv->queued_redraw = FALSE;
|
|
|
|
return;
|
|
|
|
}
|
2007-06-12 11:42:29 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* if we aren't paintable (not in a toplevel with all
|
|
|
|
* parents paintable) then do nothing.
|
|
|
|
*/
|
|
|
|
if (!CLUTTER_ACTOR_IS_MAPPED (self))
|
|
|
|
return;
|
2007-06-12 11:42:29 +00:00
|
|
|
|
2008-12-11 15:39:28 +00:00
|
|
|
/* mark that we are in the paint process */
|
|
|
|
CLUTTER_SET_PRIVATE_FLAGS (self, CLUTTER_ACTOR_IN_PAINT);
|
|
|
|
|
2007-06-12 11:42:29 +00:00
|
|
|
cogl_push_matrix();
|
|
|
|
|
2009-01-17 16:51:03 +00:00
|
|
|
if (priv->enable_model_view_transform)
|
|
|
|
_clutter_actor_apply_modelview_transform (self);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-03-19 11:38:53 +00:00
|
|
|
if (priv->has_clip)
|
2008-06-25 08:40:25 +00:00
|
|
|
{
|
2009-11-05 16:50:24 +00:00
|
|
|
cogl_clip_push_rectangle (priv->clip[0],
|
|
|
|
priv->clip[1],
|
|
|
|
priv->clip[0] + priv->clip[2],
|
|
|
|
priv->clip[1] + priv->clip[3]);
|
2008-06-25 08:40:25 +00:00
|
|
|
clip_set = TRUE;
|
|
|
|
}
|
2009-03-16 11:58:58 +00:00
|
|
|
else if (priv->clip_to_allocation)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat width, height;
|
2009-03-16 11:58:58 +00:00
|
|
|
|
|
|
|
width = priv->allocation.x2 - priv->allocation.x1;
|
|
|
|
height = priv->allocation.y2 - priv->allocation.y1;
|
|
|
|
|
2009-11-05 16:50:24 +00:00
|
|
|
cogl_clip_push_rectangle (0, 0, width, height);
|
2009-03-16 11:58:58 +00:00
|
|
|
clip_set = TRUE;
|
|
|
|
}
|
2008-03-19 11:38:53 +00:00
|
|
|
|
2008-06-05 14:32:18 +00:00
|
|
|
if (G_UNLIKELY (context->pick_mode != CLUTTER_PICK_NONE))
|
2007-05-28 18:49:34 +00:00
|
|
|
{
|
2008-06-05 14:32:18 +00:00
|
|
|
ClutterColor col = { 0, };
|
2007-11-18 23:24:55 +00:00
|
|
|
|
2009-03-30 15:41:02 +00:00
|
|
|
CLUTTER_COUNTER_INC (_clutter_uprof_context, actor_pick_counter);
|
|
|
|
|
2008-05-13 23:02:45 +00:00
|
|
|
_clutter_id_to_color (clutter_actor_get_gid (self), &col);
|
2007-11-28 12:23:31 +00:00
|
|
|
|
2007-11-18 23:24:55 +00:00
|
|
|
/* Actor will then paint silhouette of itself in supplied
|
|
|
|
* color. See clutter_stage_get_actor_at_pos() for where
|
|
|
|
* picking is enabled.
|
|
|
|
*/
|
2009-01-23 11:34:13 +00:00
|
|
|
g_signal_emit (self, actor_signals[PICK], 0, &col);
|
2007-05-28 18:49:34 +00:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2009-03-30 15:41:02 +00:00
|
|
|
CLUTTER_COUNTER_INC (_clutter_uprof_context, actor_paint_counter);
|
|
|
|
|
2007-12-03 16:29:18 +00:00
|
|
|
clutter_actor_shader_pre_paint (self, FALSE);
|
2007-12-11 14:49:41 +00:00
|
|
|
|
2009-02-17 17:22:02 +00:00
|
|
|
self->priv->queued_redraw = FALSE;
|
2008-06-05 14:32:18 +00:00
|
|
|
g_signal_emit (self, actor_signals[PAINT], 0);
|
2007-12-11 14:49:41 +00:00
|
|
|
|
2007-12-03 16:29:18 +00:00
|
|
|
clutter_actor_shader_post_paint (self);
|
2006-08-13 23:55:52 +00:00
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2008-06-25 08:40:25 +00:00
|
|
|
if (clip_set)
|
2009-02-12 11:08:00 +00:00
|
|
|
cogl_clip_pop();
|
2006-08-13 23:55:52 +00:00
|
|
|
|
2007-04-27 21:13:06 +00:00
|
|
|
cogl_pop_matrix();
|
2008-12-11 15:39:28 +00:00
|
|
|
|
|
|
|
/* paint sequence complete */
|
|
|
|
CLUTTER_UNSET_PRIVATE_FLAGS (self, CLUTTER_ACTOR_IN_PAINT);
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
/* internal helper function set the rotation angle without affecting
|
|
|
|
the center point
|
2008-04-23 12:39:50 +00:00
|
|
|
*/
|
2009-01-23 15:55:41 +00:00
|
|
|
static void
|
2008-04-23 12:39:50 +00:00
|
|
|
clutter_actor_set_rotation_internal (ClutterActor *self,
|
|
|
|
ClutterRotateAxis axis,
|
2009-01-23 15:55:41 +00:00
|
|
|
gdouble angle)
|
2008-04-23 12:39:50 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv = self->priv;
|
|
|
|
|
|
|
|
g_object_ref (self);
|
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
switch (axis)
|
|
|
|
{
|
|
|
|
case CLUTTER_X_AXIS:
|
|
|
|
priv->rxang = angle;
|
|
|
|
g_object_notify (G_OBJECT (self), "rotation-angle-x");
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_Y_AXIS:
|
|
|
|
priv->ryang = angle;
|
|
|
|
g_object_notify (G_OBJECT (self), "rotation-angle-y");
|
|
|
|
break;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2008-04-23 12:39:50 +00:00
|
|
|
case CLUTTER_Z_AXIS:
|
|
|
|
priv->rzang = angle;
|
|
|
|
g_object_notify (G_OBJECT (self), "rotation-angle-z");
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
g_object_unref (self);
|
|
|
|
|
|
|
|
if (CLUTTER_ACTOR_IS_VISIBLE (self))
|
|
|
|
clutter_actor_queue_redraw (self);
|
|
|
|
}
|
|
|
|
|
2007-10-12 08:17:00 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_set_property (GObject *object,
|
2007-06-14 15:29:52 +00:00
|
|
|
guint prop_id,
|
2007-10-12 08:17:00 +00:00
|
|
|
const GValue *value,
|
2007-06-14 15:29:52 +00:00
|
|
|
GParamSpec *pspec)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2009-05-31 14:15:46 +00:00
|
|
|
ClutterActor *actor = CLUTTER_ACTOR (object);
|
|
|
|
ClutterActorPrivate *priv = actor->priv;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-10-12 08:17:00 +00:00
|
|
|
switch (prop_id)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
|
|
|
case PROP_X:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_x (actor, g_value_get_float (value));
|
2006-06-13 13:17:45 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_Y:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_y (actor, g_value_get_float (value));
|
2006-06-13 13:17:45 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_WIDTH:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_width (actor, g_value_get_float (value));
|
2006-06-13 13:17:45 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_HEIGHT:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_height (actor, g_value_get_float (value));
|
2006-06-13 13:17:45 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_FIXED_X:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_x (actor, g_value_get_float (value));
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_FIXED_Y:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_y (actor, g_value_get_float (value));
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_FIXED_POSITION_SET:
|
|
|
|
clutter_actor_set_fixed_position_set (actor, g_value_get_boolean (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_MIN_WIDTH:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_min_width (actor, g_value_get_float (value));
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_MIN_HEIGHT:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_min_height (actor, g_value_get_float (value));
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_NATURAL_WIDTH:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_natural_width (actor, g_value_get_float (value));
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_NATURAL_HEIGHT:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_natural_height (actor, g_value_get_float (value));
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_MIN_WIDTH_SET:
|
|
|
|
clutter_actor_set_min_width_set (actor, g_value_get_boolean (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_MIN_HEIGHT_SET:
|
|
|
|
clutter_actor_set_min_height_set (actor, g_value_get_boolean (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_NATURAL_WIDTH_SET:
|
|
|
|
clutter_actor_set_natural_width_set (actor, g_value_get_boolean (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_NATURAL_HEIGHT_SET:
|
|
|
|
clutter_actor_set_natural_height_set (actor, g_value_get_boolean (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_REQUEST_MODE:
|
|
|
|
clutter_actor_set_request_mode (actor, g_value_get_enum (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-10-12 12:42:54 +00:00
|
|
|
case PROP_DEPTH:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_depth (actor, g_value_get_float (value));
|
2007-10-12 12:42:54 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_OPACITY:
|
|
|
|
clutter_actor_set_opacity (actor, g_value_get_uchar (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_NAME:
|
|
|
|
clutter_actor_set_name (actor, g_value_get_string (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
case PROP_VISIBLE:
|
|
|
|
if (g_value_get_boolean (value) == TRUE)
|
|
|
|
clutter_actor_show (actor);
|
|
|
|
else
|
|
|
|
clutter_actor_hide (actor);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
case PROP_SCALE_X:
|
2009-01-23 11:11:24 +00:00
|
|
|
clutter_actor_set_scale (actor,
|
|
|
|
g_value_get_double (value),
|
|
|
|
priv->scale_y);
|
2007-09-27 21:38:38 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
case PROP_SCALE_Y:
|
2009-01-23 11:11:24 +00:00
|
|
|
clutter_actor_set_scale (actor,
|
|
|
|
priv->scale_x,
|
|
|
|
g_value_get_double (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
case PROP_SCALE_CENTER_X:
|
|
|
|
{
|
2009-05-31 14:15:46 +00:00
|
|
|
gfloat center_x = g_value_get_float (value);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat center_y;
|
2009-01-23 11:11:24 +00:00
|
|
|
|
|
|
|
clutter_anchor_coord_get_units (actor, &priv->scale_center,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
NULL,
|
|
|
|
¢er_y,
|
|
|
|
NULL);
|
|
|
|
clutter_actor_set_scale_full (actor,
|
|
|
|
priv->scale_x,
|
|
|
|
priv->scale_y,
|
|
|
|
center_x,
|
|
|
|
center_y);
|
2009-01-23 11:11:24 +00:00
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
case PROP_SCALE_CENTER_Y:
|
|
|
|
{
|
2009-05-31 14:15:46 +00:00
|
|
|
gfloat center_y = g_value_get_float (value);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat center_x;
|
2009-01-23 11:11:24 +00:00
|
|
|
|
|
|
|
clutter_anchor_coord_get_units (actor, &priv->scale_center,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
¢er_x,
|
|
|
|
NULL,
|
|
|
|
NULL);
|
|
|
|
clutter_actor_set_scale_full (actor,
|
|
|
|
priv->scale_x,
|
|
|
|
priv->scale_y,
|
|
|
|
center_x,
|
|
|
|
center_y);
|
2009-01-23 11:11:24 +00:00
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
case PROP_SCALE_GRAVITY:
|
|
|
|
clutter_actor_set_scale_with_gravity (actor,
|
|
|
|
priv->scale_x,
|
|
|
|
priv->scale_y,
|
|
|
|
g_value_get_enum (value));
|
2007-09-27 21:38:38 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-07-06 18:12:19 +00:00
|
|
|
case PROP_CLIP:
|
|
|
|
{
|
2009-05-31 14:15:46 +00:00
|
|
|
const ClutterGeometry *geom = g_value_get_boxed (value);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-07-06 18:12:19 +00:00
|
|
|
clutter_actor_set_clip (actor,
|
|
|
|
geom->x, geom->y,
|
|
|
|
geom->width, geom->height);
|
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-03-16 11:58:58 +00:00
|
|
|
case PROP_CLIP_TO_ALLOCATION:
|
|
|
|
if (priv->clip_to_allocation != g_value_get_boolean (value))
|
|
|
|
{
|
|
|
|
priv->clip_to_allocation = g_value_get_boolean (value);
|
|
|
|
clutter_actor_queue_redraw (actor);
|
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-10-17 16:43:46 +00:00
|
|
|
case PROP_REACTIVE:
|
2007-11-15 17:35:47 +00:00
|
|
|
clutter_actor_set_reactive (actor, g_value_get_boolean (value));
|
2007-10-17 16:43:46 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_ANGLE_X:
|
2009-01-23 15:55:41 +00:00
|
|
|
clutter_actor_set_rotation_internal (actor,
|
|
|
|
CLUTTER_X_AXIS,
|
|
|
|
g_value_get_double (value));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_ANGLE_Y:
|
2009-01-23 15:55:41 +00:00
|
|
|
clutter_actor_set_rotation_internal (actor,
|
|
|
|
CLUTTER_Y_AXIS,
|
|
|
|
g_value_get_double (value));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_ANGLE_Z:
|
2009-01-23 15:55:41 +00:00
|
|
|
clutter_actor_set_rotation_internal (actor,
|
|
|
|
CLUTTER_Z_AXIS,
|
|
|
|
g_value_get_double (value));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_CENTER_X:
|
|
|
|
{
|
2009-01-23 15:55:41 +00:00
|
|
|
const ClutterVertex *center;
|
|
|
|
|
|
|
|
if ((center = g_value_get_boxed (value)))
|
|
|
|
clutter_actor_set_rotation (actor,
|
|
|
|
CLUTTER_X_AXIS,
|
|
|
|
priv->rxang,
|
|
|
|
center->x,
|
|
|
|
center->y,
|
|
|
|
center->z);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_CENTER_Y:
|
|
|
|
{
|
2009-01-23 15:55:41 +00:00
|
|
|
const ClutterVertex *center;
|
|
|
|
|
|
|
|
if ((center = g_value_get_boxed (value)))
|
|
|
|
clutter_actor_set_rotation (actor,
|
|
|
|
CLUTTER_Y_AXIS,
|
|
|
|
priv->ryang,
|
|
|
|
center->x,
|
|
|
|
center->y,
|
|
|
|
center->z);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_CENTER_Z:
|
|
|
|
{
|
2009-01-23 15:55:41 +00:00
|
|
|
const ClutterVertex *center;
|
|
|
|
|
|
|
|
if ((center = g_value_get_boxed (value)))
|
|
|
|
clutter_actor_set_rotation (actor,
|
|
|
|
CLUTTER_Z_AXIS,
|
|
|
|
priv->rzang,
|
|
|
|
center->x,
|
|
|
|
center->y,
|
|
|
|
center->z);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
case PROP_ROTATION_CENTER_Z_GRAVITY:
|
2009-05-31 14:15:46 +00:00
|
|
|
clutter_actor_set_z_rotation_from_gravity (actor, priv->rzang,
|
|
|
|
g_value_get_enum (value));
|
2009-01-23 15:55:41 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-03-06 04:37:00 +00:00
|
|
|
case PROP_ANCHOR_X:
|
2008-09-10 11:20:22 +00:00
|
|
|
{
|
2009-05-31 14:15:46 +00:00
|
|
|
gfloat anchor_x = g_value_get_float (value);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat anchor_y;
|
2009-01-22 13:07:33 +00:00
|
|
|
|
|
|
|
clutter_anchor_coord_get_units (actor, &priv->anchor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
NULL,
|
|
|
|
&anchor_y,
|
|
|
|
NULL);
|
|
|
|
clutter_actor_set_anchor_point (actor, anchor_x, anchor_y);
|
2008-09-10 11:20:22 +00:00
|
|
|
}
|
2008-03-06 04:37:00 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-03-06 04:37:00 +00:00
|
|
|
case PROP_ANCHOR_Y:
|
2008-09-10 11:20:22 +00:00
|
|
|
{
|
2009-06-03 09:09:56 +00:00
|
|
|
gfloat anchor_y = g_value_get_float (value);
|
2009-05-31 14:15:46 +00:00
|
|
|
gfloat anchor_x;
|
2009-01-22 13:07:33 +00:00
|
|
|
|
|
|
|
clutter_anchor_coord_get_units (actor, &priv->anchor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&anchor_x,
|
|
|
|
NULL,
|
|
|
|
NULL);
|
|
|
|
clutter_actor_set_anchor_point (actor, anchor_x, anchor_y);
|
2008-09-10 11:20:22 +00:00
|
|
|
}
|
2008-03-06 04:37:00 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
case PROP_ANCHOR_GRAVITY:
|
|
|
|
clutter_actor_set_anchor_point_from_gravity (actor,
|
|
|
|
g_value_get_enum (value));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-04-04 13:20:02 +00:00
|
|
|
case PROP_SHOW_ON_SET_PARENT:
|
|
|
|
priv->show_on_set_parent = g_value_get_boolean (value);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-11-06 16:55:46 +00:00
|
|
|
case PROP_TEXT_DIRECTION:
|
|
|
|
clutter_actor_set_text_direction (actor, g_value_get_enum (value));
|
|
|
|
break;
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
default:
|
|
|
|
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-10-12 08:17:00 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_get_property (GObject *object,
|
2008-02-04 22:35:38 +00:00
|
|
|
guint prop_id,
|
|
|
|
GValue *value,
|
|
|
|
GParamSpec *pspec)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2009-05-31 14:15:46 +00:00
|
|
|
ClutterActor *actor = CLUTTER_ACTOR (object);
|
|
|
|
ClutterActorPrivate *priv = actor->priv;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-10-12 08:17:00 +00:00
|
|
|
switch (prop_id)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
|
|
|
case PROP_X:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, clutter_actor_get_x (actor));
|
2006-06-13 13:17:45 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_Y:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, clutter_actor_get_y (actor));
|
2006-06-13 13:17:45 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_WIDTH:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, clutter_actor_get_width (actor));
|
2006-06-13 13:17:45 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_HEIGHT:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, clutter_actor_get_height (actor));
|
2006-06-13 13:17:45 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_FIXED_X:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, priv->fixed_x);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_FIXED_Y:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, priv->fixed_y);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_FIXED_POSITION_SET:
|
|
|
|
g_value_set_boolean (value, priv->position_set);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_MIN_WIDTH:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, priv->request_min_width);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_MIN_HEIGHT:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, priv->request_min_height);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_NATURAL_WIDTH:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, priv->request_natural_width);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_NATURAL_HEIGHT:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, priv->request_natural_height);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_MIN_WIDTH_SET:
|
|
|
|
g_value_set_boolean (value, priv->min_width_set);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_MIN_HEIGHT_SET:
|
|
|
|
g_value_set_boolean (value, priv->min_height_set);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_NATURAL_WIDTH_SET:
|
|
|
|
g_value_set_boolean (value, priv->natural_width_set);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_NATURAL_HEIGHT_SET:
|
|
|
|
g_value_set_boolean (value, priv->natural_height_set);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_REQUEST_MODE:
|
|
|
|
g_value_set_enum (value, priv->request_mode);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
case PROP_ALLOCATION:
|
|
|
|
g_value_set_boxed (value, &priv->allocation);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-10-12 12:42:54 +00:00
|
|
|
case PROP_DEPTH:
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, clutter_actor_get_depth (actor));
|
2007-10-12 12:42:54 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_OPACITY:
|
|
|
|
g_value_set_uchar (value, priv->opacity);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
case PROP_NAME:
|
|
|
|
g_value_set_string (value, priv->name);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
case PROP_VISIBLE:
|
2009-05-05 14:36:29 +00:00
|
|
|
g_value_set_boolean (value, CLUTTER_ACTOR_IS_VISIBLE (actor));
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
case PROP_MAPPED:
|
2009-05-05 14:36:29 +00:00
|
|
|
g_value_set_boolean (value, CLUTTER_ACTOR_IS_MAPPED (actor));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-05-05 14:36:29 +00:00
|
|
|
case PROP_REALIZED:
|
|
|
|
g_value_set_boolean (value, CLUTTER_ACTOR_IS_REALIZED (actor));
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-07-06 18:12:19 +00:00
|
|
|
case PROP_HAS_CLIP:
|
|
|
|
g_value_set_boolean (value, priv->has_clip);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2006-07-06 18:12:19 +00:00
|
|
|
case PROP_CLIP:
|
2007-12-24 12:53:04 +00:00
|
|
|
{
|
|
|
|
ClutterGeometry clip = { 0, };
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clip.x = priv->clip[0];
|
|
|
|
clip.y = priv->clip[1];
|
|
|
|
clip.width = priv->clip[2];
|
|
|
|
clip.height = priv->clip[3];
|
2007-12-24 12:53:04 +00:00
|
|
|
|
|
|
|
g_value_set_boxed (value, &clip);
|
|
|
|
}
|
2006-07-06 18:12:19 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-03-16 11:58:58 +00:00
|
|
|
case PROP_CLIP_TO_ALLOCATION:
|
|
|
|
g_value_set_boolean (value, priv->clip_to_allocation);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
case PROP_SCALE_X:
|
2009-01-23 11:11:24 +00:00
|
|
|
g_value_set_double (value, priv->scale_x);
|
2007-09-27 21:38:38 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
case PROP_SCALE_Y:
|
2009-01-23 11:11:24 +00:00
|
|
|
g_value_set_double (value, priv->scale_y);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
case PROP_SCALE_CENTER_X:
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat center;
|
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
clutter_actor_get_scale_center (actor, ¢er, NULL);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, center);
|
2009-01-23 11:11:24 +00:00
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
case PROP_SCALE_CENTER_Y:
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat center;
|
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
clutter_actor_get_scale_center (actor, NULL, ¢er);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, center);
|
2009-01-23 11:11:24 +00:00
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
case PROP_SCALE_GRAVITY:
|
|
|
|
g_value_set_enum (value, clutter_actor_get_scale_gravity (actor));
|
2007-09-27 21:38:38 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-10-17 16:43:46 +00:00
|
|
|
case PROP_REACTIVE:
|
2008-02-01 11:22:59 +00:00
|
|
|
g_value_set_boolean (value, clutter_actor_get_reactive (actor));
|
2007-10-17 16:43:46 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_ANGLE_X:
|
2009-01-23 15:55:41 +00:00
|
|
|
g_value_set_double (value, priv->rxang);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_ANGLE_Y:
|
2009-01-23 15:55:41 +00:00
|
|
|
g_value_set_double (value, priv->ryang);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_ANGLE_Z:
|
2009-01-23 15:55:41 +00:00
|
|
|
g_value_set_double (value, priv->rzang);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_CENTER_X:
|
|
|
|
{
|
2009-01-23 15:55:41 +00:00
|
|
|
ClutterVertex center;
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_get_rotation (actor, CLUTTER_X_AXIS,
|
|
|
|
¢er.x,
|
|
|
|
¢er.y,
|
|
|
|
¢er.z);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
|
|
|
g_value_set_boxed (value, ¢er);
|
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_CENTER_Y:
|
|
|
|
{
|
2009-01-23 15:55:41 +00:00
|
|
|
ClutterVertex center;
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_get_rotation (actor, CLUTTER_Y_AXIS,
|
|
|
|
¢er.x,
|
|
|
|
¢er.y,
|
|
|
|
¢er.z);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
|
|
|
g_value_set_boxed (value, ¢er);
|
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
case PROP_ROTATION_CENTER_Z:
|
|
|
|
{
|
2009-01-23 15:55:41 +00:00
|
|
|
ClutterVertex center;
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_get_rotation (actor, CLUTTER_Z_AXIS,
|
|
|
|
¢er.x,
|
|
|
|
¢er.y,
|
|
|
|
¢er.z);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
|
|
|
g_value_set_boxed (value, ¢er);
|
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
case PROP_ROTATION_CENTER_Z_GRAVITY:
|
|
|
|
g_value_set_enum (value, clutter_actor_get_z_rotation_gravity (actor));
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-03-06 04:37:00 +00:00
|
|
|
case PROP_ANCHOR_X:
|
2009-01-22 13:07:33 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat anchor_x;
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_anchor_coord_get_units (actor, &priv->anchor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&anchor_x,
|
|
|
|
NULL,
|
|
|
|
NULL);
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, anchor_x);
|
2009-01-22 13:07:33 +00:00
|
|
|
}
|
2008-03-06 04:37:00 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-03-06 04:37:00 +00:00
|
|
|
case PROP_ANCHOR_Y:
|
2009-01-22 13:07:33 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat anchor_y;
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_anchor_coord_get_units (actor, &priv->anchor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
NULL,
|
|
|
|
&anchor_y,
|
|
|
|
NULL);
|
2009-05-31 14:15:46 +00:00
|
|
|
g_value_set_float (value, anchor_y);
|
2009-01-22 13:07:33 +00:00
|
|
|
}
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
case PROP_ANCHOR_GRAVITY:
|
|
|
|
g_value_set_enum (value, clutter_actor_get_anchor_point_gravity (actor));
|
2008-03-06 04:37:00 +00:00
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2008-04-04 13:20:02 +00:00
|
|
|
case PROP_SHOW_ON_SET_PARENT:
|
|
|
|
g_value_set_boolean (value, priv->show_on_set_parent);
|
|
|
|
break;
|
2009-05-31 14:15:46 +00:00
|
|
|
|
2009-11-06 16:55:46 +00:00
|
|
|
case PROP_TEXT_DIRECTION:
|
|
|
|
g_value_set_enum (value, priv->text_direction);
|
|
|
|
break;
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
default:
|
|
|
|
G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-10-12 08:17:00 +00:00
|
|
|
static void
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_dispose (GObject *object)
|
|
|
|
{
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
ClutterActor *self = CLUTTER_ACTOR (object);
|
2008-06-13 13:59:07 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2009-03-17 14:12:01 +00:00
|
|
|
CLUTTER_NOTE (MISC, "Disposing of object (id=%d) of type '%s' (ref_count:%d)",
|
2006-11-22 20:52:27 +00:00
|
|
|
self->priv->id,
|
2007-11-27 11:05:26 +00:00
|
|
|
g_type_name (G_OBJECT_TYPE (self)),
|
|
|
|
object->ref_count);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-06-13 13:59:07 +00:00
|
|
|
/* avoid recursing when called from clutter_actor_destroy() */
|
2009-06-03 10:12:09 +00:00
|
|
|
if (priv->parent_actor != NULL)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2008-06-13 13:59:07 +00:00
|
|
|
ClutterActor *parent = priv->parent_actor;
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
|
actor: Add internal child flag
ClutterActor checks, when destroying and reparenting, if the parent
actor implements the Container interface, and automatically calls the
remove() method to perform a clean removal.
Actors implementing Container, though, might have internal children;
that is, children that are not added through the Container API. It is
already possible to iterate through them using the Container API to
avoid breaking invariants - but calling clutter_actor_destroy() on
these children (even from the Container implementation, and thus outside
of Clutter's control) will either lead to leaks or to segmentation
faults.
Clutter needs a way to distinguish a clutter_actor_set_parent() done on
an internal child from one done on a "public" child; for this reason, a
push/pop pair of functions should be available to Actor implementations
to mark the section where they wish to add internal children:
➔ clutter_actor_push_internal ();
...
clutter_actor_set_parent (child1, parent);
clutter_actor_set_parent (child2, parent);
...
➔ clutter_actor_pop_internal ();
The set_parent() call will automatically set the newly added
INTERNAL_CHILD private flag on each child, and both
clutter_actor_destroy() and clutter_actor_unparent() will check for the
flag before deciding whether to call the Container's remove method.
2009-12-18 23:20:04 +00:00
|
|
|
/* go through the Container implementation unless this
|
|
|
|
* is an internal child and has been marked as such
|
|
|
|
*/
|
|
|
|
if (CLUTTER_IS_CONTAINER (parent) &&
|
|
|
|
!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_INTERNAL_CHILD))
|
|
|
|
{
|
|
|
|
clutter_container_remove_actor (CLUTTER_CONTAINER (parent), self);
|
|
|
|
}
|
2008-06-13 13:59:07 +00:00
|
|
|
else
|
|
|
|
priv->parent_actor = NULL;
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
}
|
2007-10-12 08:17:00 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* parent should be gone */
|
|
|
|
g_assert (priv->parent_actor == NULL);
|
2009-05-08 16:14:49 +00:00
|
|
|
|
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL))
|
|
|
|
{
|
|
|
|
/* can't be mapped or realized with no parent */
|
|
|
|
g_assert (!CLUTTER_ACTOR_IS_MAPPED (self));
|
|
|
|
g_assert (!CLUTTER_ACTOR_IS_REALIZED (self));
|
|
|
|
}
|
2008-06-13 14:16:22 +00:00
|
|
|
|
2008-06-13 13:59:07 +00:00
|
|
|
destroy_shader_data (self);
|
|
|
|
|
2008-12-23 14:15:10 +00:00
|
|
|
if (priv->pango_context)
|
|
|
|
{
|
|
|
|
g_object_unref (priv->pango_context);
|
|
|
|
priv->pango_context = NULL;
|
|
|
|
}
|
|
|
|
|
2008-06-13 13:59:07 +00:00
|
|
|
g_signal_emit (self, actor_signals[DESTROY], 0);
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
G_OBJECT_CLASS (clutter_actor_parent_class)->dispose (object);
|
|
|
|
}
|
|
|
|
|
2007-10-12 08:17:00 +00:00
|
|
|
static void
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_finalize (GObject *object)
|
|
|
|
{
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
ClutterActor *actor = CLUTTER_ACTOR (object);
|
|
|
|
|
2009-03-17 14:12:01 +00:00
|
|
|
CLUTTER_NOTE (MISC, "Finalize object (id=%d) of type '%s'",
|
2007-11-27 11:05:26 +00:00
|
|
|
actor->priv->id,
|
|
|
|
g_type_name (G_OBJECT_TYPE (actor)));
|
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
g_free (actor->priv->name);
|
2008-03-07 17:34:40 +00:00
|
|
|
clutter_id_pool_remove (CLUTTER_CONTEXT()->id_pool, actor->priv->id);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
G_OBJECT_CLASS (clutter_actor_parent_class)->finalize (object);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_class_init (ClutterActorClass *klass)
|
|
|
|
{
|
|
|
|
GObjectClass *object_class = G_OBJECT_CLASS (klass);
|
2009-01-22 13:07:33 +00:00
|
|
|
GParamSpec *pspec;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
|
|
|
object_class->set_property = clutter_actor_set_property;
|
|
|
|
object_class->get_property = clutter_actor_get_property;
|
|
|
|
object_class->dispose = clutter_actor_dispose;
|
|
|
|
object_class->finalize = clutter_actor_finalize;
|
|
|
|
|
|
|
|
g_type_class_add_private (klass, sizeof (ClutterActorPrivate));
|
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:x:
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* X coordinate of the actor in pixels. If written, forces a fixed
|
|
|
|
* position for the actor. If read, returns the fixed position if any,
|
|
|
|
* otherwise the allocation if available, otherwise 0.
|
2006-11-20 12:02:24 +00:00
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("x",
|
|
|
|
"X coordinate",
|
|
|
|
"X coordinate of the actor",
|
|
|
|
-G_MAXFLOAT, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_X, pspec);
|
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:y:
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Y coordinate of the actor in pixels. If written, forces a fixed
|
|
|
|
* position for the actor. If read, returns the fixed position if
|
|
|
|
* any, otherwise the allocation if available, otherwise 0.
|
2006-11-20 12:02:24 +00:00
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("y",
|
|
|
|
"Y coordinate",
|
|
|
|
"Y coordinate of the actor",
|
|
|
|
-G_MAXFLOAT, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_Y, pspec);
|
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:width:
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Width of the actor (in pixels). If written, forces the minimum and
|
|
|
|
* natural size request of the actor to the given width. If read, returns
|
|
|
|
* the allocated width if available, otherwise the width request.
|
2006-11-20 12:02:24 +00:00
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("width",
|
|
|
|
"Width",
|
|
|
|
"Width of the actor",
|
|
|
|
0.0, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_WIDTH, pspec);
|
2006-11-20 12:02:24 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:height:
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Height of the actor (in pixels). If written, forces the minimum and
|
|
|
|
* natural size request of the actor to the given height. If read, returns
|
|
|
|
* the allocated height if available, otherwise the height request.
|
2006-11-20 12:02:24 +00:00
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("height",
|
|
|
|
"Height",
|
|
|
|
"Height of the actor",
|
|
|
|
0.0, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_HEIGHT, pspec);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2007-10-12 12:42:54 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:fixed-x:
|
2007-10-12 12:42:54 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The fixed X position of the actor in pixels.
|
|
|
|
*
|
|
|
|
* Writing this property sets #ClutterActor:fixed-position-set
|
|
|
|
* property as well, as a side effect
|
2007-10-12 12:42:54 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Since: 0.8
|
2007-10-12 12:42:54 +00:00
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("fixed-x",
|
|
|
|
"Fixed X",
|
|
|
|
"Forced X position of the actor",
|
|
|
|
-G_MAXFLOAT, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_FIXED_X, pspec);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:fixed-y:
|
|
|
|
*
|
|
|
|
* The fixed Y position of the actor in pixels.
|
2006-11-20 12:02:24 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Writing this property sets the #ClutterActor:fixed-position-set
|
|
|
|
* property as well, as a side effect
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
2006-11-20 12:02:24 +00:00
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("fixed-y",
|
|
|
|
"Fixed Y",
|
|
|
|
"Forced Y position of the actor",
|
|
|
|
-G_MAXFLOAT, G_MAXFLOAT,
|
|
|
|
0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_FIXED_Y, pspec);
|
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:fixed-position-set:
|
2006-11-20 12:02:24 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* This flag controls whether the #ClutterActor:fixed-x and
|
|
|
|
* #ClutterActor:fixed-y properties are used
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
2006-11-20 12:02:24 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boolean ("fixed-position-set",
|
|
|
|
"Fixed position set",
|
|
|
|
"Whether to use fixed positioning "
|
|
|
|
"for the actor",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_FIXED_POSITION_SET,
|
|
|
|
pspec);
|
|
|
|
|
2007-10-17 16:43:46 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:min-width:
|
|
|
|
*
|
|
|
|
* A forced minimum width request for the actor, in pixels
|
|
|
|
*
|
|
|
|
* Writing this property sets the #ClutterActor:min-width-set property
|
|
|
|
* as well, as a side effect.
|
2007-10-17 16:43:46 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*This property overrides the usual width request of the actor.
|
2007-10-17 16:43:46 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Since: 0.8
|
2007-10-17 16:43:46 +00:00
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("min-width",
|
|
|
|
"Min Width",
|
|
|
|
"Forced minimum width request for the actor",
|
|
|
|
0.0, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_MIN_WIDTH, pspec);
|
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:min-height:
|
|
|
|
*
|
|
|
|
* A forced minimum height request for the actor, in pixels
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Writing this property sets the #ClutterActor:min-height-set property
|
|
|
|
* as well, as a side effect. This property overrides the usual height
|
|
|
|
* request of the actor.
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("min-height",
|
|
|
|
"Min Height",
|
|
|
|
"Forced minimum height request for the actor",
|
|
|
|
0.0, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_MIN_HEIGHT, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:natural-width:
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* A forced natural width request for the actor, in pixels
|
|
|
|
*
|
|
|
|
* Writing this property sets the #ClutterActor:natural-width-set
|
|
|
|
* property as well, as a side effect. This property overrides the
|
|
|
|
* usual width request of the actor
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("natural-width",
|
|
|
|
"Natural Width",
|
|
|
|
"Forced natural width request for the actor",
|
|
|
|
0.0, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_NATURAL_WIDTH, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:natural-height:
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* A forced natural height request for the actor, in pixels
|
|
|
|
*
|
|
|
|
* Writing this property sets the #ClutterActor:natural-height-set
|
|
|
|
* property as well, as a side effect. This property overrides the
|
|
|
|
* usual height request of the actor
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("natural-height",
|
|
|
|
"Natural Height",
|
|
|
|
"Forced natural height request for the actor",
|
|
|
|
0.0, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_NATURAL_HEIGHT, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:min-width-set:
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* This flag controls whether the #ClutterActor:min-width property
|
|
|
|
* is used
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boolean ("min-width-set",
|
|
|
|
"Minimum width set",
|
|
|
|
"Whether to use the min-width property",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_MIN_WIDTH_SET, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:min-height-set:
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* This flag controls whether the #ClutterActor:min-height property
|
|
|
|
* is used
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boolean ("min-height-set",
|
|
|
|
"Minimum height set",
|
|
|
|
"Whether to use the min-height property",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_MIN_HEIGHT_SET, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:natural-width-set:
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* This flag controls whether the #ClutterActor:natural-width property
|
|
|
|
* is used
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boolean ("natural-width-set",
|
|
|
|
"Natural width set",
|
|
|
|
"Whether to use the natural-width property",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_NATURAL_WIDTH_SET,
|
|
|
|
pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* ClutterActor:natural-height-set:
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* This flag controls whether the #ClutterActor:natural-height property
|
|
|
|
* is used
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boolean ("natural-height-set",
|
|
|
|
"Natural height set",
|
|
|
|
"Whether to use the natural-height property",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_NATURAL_HEIGHT_SET,
|
|
|
|
pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:allocation:
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The allocation for the actor, in pixels
|
|
|
|
*
|
|
|
|
* This is property is read-only, but you might monitor it to know when an
|
|
|
|
* actor moves or resizes
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boxed ("allocation",
|
|
|
|
"Allocation",
|
|
|
|
"The actor's allocation",
|
|
|
|
CLUTTER_TYPE_ACTOR_BOX,
|
|
|
|
CLUTTER_PARAM_READABLE);
|
|
|
|
g_object_class_install_property (object_class, PROP_ALLOCATION, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:request-mode:
|
|
|
|
*
|
|
|
|
* Request mode for the #ClutterActor. The request mode determines the
|
|
|
|
* type of geometry management used by the actor, either height for width
|
|
|
|
* (the default) or width for height.
|
|
|
|
*
|
|
|
|
* For actors implementing height for width, the parent container should get
|
|
|
|
* the preferred width first, and then the preferred height for that width.
|
|
|
|
*
|
|
|
|
* For actors implementing width for height, the parent container should get
|
|
|
|
* the preferred height first, and then the preferred width for that height.
|
|
|
|
*
|
|
|
|
* For instance:
|
|
|
|
*
|
|
|
|
* |[
|
|
|
|
* ClutterRequestMode mode;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* gfloat natural_width, min_width;
|
|
|
|
* gfloat natural_height, min_height;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
2009-10-07 14:15:02 +00:00
|
|
|
* mode = clutter_actor_get_request_mode (child);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* if (mode == CLUTTER_REQUEST_HEIGHT_FOR_WIDTH)
|
|
|
|
* {
|
|
|
|
* clutter_actor_get_preferred_width (child, -1,
|
|
|
|
* &min_width,
|
|
|
|
* &natural_width);
|
|
|
|
* clutter_actor_get_preferred_height (child, natural_width,
|
|
|
|
* &min_height,
|
|
|
|
* &natural_height);
|
|
|
|
* }
|
|
|
|
* else
|
|
|
|
* {
|
|
|
|
* clutter_actor_get_preferred_height (child, -1,
|
|
|
|
* &min_height,
|
|
|
|
* &natural_height);
|
|
|
|
* clutter_actor_get_preferred_width (child, natural_height,
|
|
|
|
* &min_width,
|
|
|
|
* &natural_width);
|
|
|
|
* }
|
|
|
|
* ]|
|
|
|
|
*
|
|
|
|
* will retrieve the minimum and natural width and height depending on the
|
|
|
|
* preferred request mode of the #ClutterActor "child".
|
|
|
|
*
|
|
|
|
* The clutter_actor_get_preferred_size() function will implement this
|
|
|
|
* check for you.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_enum ("request-mode",
|
|
|
|
"Request Mode",
|
|
|
|
"The actor's request mode",
|
|
|
|
CLUTTER_TYPE_REQUEST_MODE,
|
|
|
|
CLUTTER_REQUEST_HEIGHT_FOR_WIDTH,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_REQUEST_MODE, pspec);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* ClutterActor:depth:
|
|
|
|
*
|
2009-05-31 14:15:46 +00:00
|
|
|
* The position of the actor on the Z axis
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("depth",
|
|
|
|
"Depth",
|
|
|
|
"Position on the Z axis",
|
|
|
|
-G_MAXFLOAT, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_DEPTH, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:opacity:
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Opacity of the actor, between 0 (fully transparent) and
|
|
|
|
* 255 (fully opaque)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_uchar ("opacity",
|
|
|
|
"Opacity",
|
|
|
|
"Opacity of actor",
|
2009-05-31 14:15:46 +00:00
|
|
|
0, 255,
|
|
|
|
255,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_OPACITY, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:visible:
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Whether the actor is set to be visible or not
|
|
|
|
*
|
|
|
|
* See also #ClutterActor:mapped
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boolean ("visible",
|
|
|
|
"Visible",
|
|
|
|
"Whether the actor is visible or not",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_VISIBLE, pspec);
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* ClutterActor:mapped:
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Whether the actor is mapped (will be painted when the stage
|
|
|
|
* to which it belongs is mapped)
|
2009-05-05 14:36:29 +00:00
|
|
|
*
|
|
|
|
* Since: 1.0
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
*/
|
2009-05-05 14:36:29 +00:00
|
|
|
pspec = g_param_spec_boolean ("mapped",
|
|
|
|
"Mapped",
|
|
|
|
"Whether the actor will be painted",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READABLE);
|
|
|
|
g_object_class_install_property (object_class, PROP_MAPPED, pspec);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* ClutterActor:realized:
|
|
|
|
*
|
|
|
|
* Whether the actor has been realized
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
pspec = g_param_spec_boolean ("realized",
|
|
|
|
"Realized",
|
|
|
|
"Whether the actor has been realized",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READABLE);
|
|
|
|
g_object_class_install_property (object_class, PROP_REALIZED, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:reactive:
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Whether the actor is reactive to events or not
|
|
|
|
*
|
|
|
|
* Only reactive actors will emit event-related signals
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boolean ("reactive",
|
|
|
|
"Reactive",
|
|
|
|
"Whether the actor is reactive to events",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_REACTIVE, pspec);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:has-clip:
|
2006-11-20 12:02:24 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Whether the actor has the #ClutterActor:clip property set or not
|
2006-11-20 12:02:24 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boolean ("has-clip",
|
|
|
|
"Has Clip",
|
|
|
|
"Whether the actor has a clip set",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READABLE);
|
|
|
|
g_object_class_install_property (object_class, PROP_HAS_CLIP, pspec);
|
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:clip:
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The clip region for the actor, in actor-relative coordinates
|
|
|
|
*
|
|
|
|
* Every part of the actor outside the clip region will not be
|
|
|
|
* painted
|
2006-11-20 12:02:24 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boxed ("clip",
|
|
|
|
"Clip",
|
|
|
|
"The clip region for the actor",
|
|
|
|
CLUTTER_TYPE_GEOMETRY,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_CLIP, pspec);
|
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:name:
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The name of the actor
|
2006-11-20 12:02:24 +00:00
|
|
|
*
|
|
|
|
* Since: 0.2
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_string ("name",
|
|
|
|
"Name",
|
|
|
|
"Name of the actor",
|
|
|
|
NULL,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_NAME, pspec);
|
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:scale-x:
|
2007-09-27 21:38:38 +00:00
|
|
|
*
|
|
|
|
* The horizontal scale of the actor
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_double ("scale-x",
|
|
|
|
"Scale X",
|
|
|
|
"Scale factor on the X axis",
|
|
|
|
0.0, G_MAXDOUBLE,
|
|
|
|
1.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_SCALE_X, pspec);
|
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:scale-y:
|
2007-09-27 21:38:38 +00:00
|
|
|
*
|
|
|
|
* The vertical scale of the actor
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_double ("scale-y",
|
|
|
|
"Scale Y",
|
|
|
|
"Scale factor on the Y axis",
|
|
|
|
0.0, G_MAXDOUBLE,
|
|
|
|
1.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_SCALE_Y, pspec);
|
2009-01-23 11:11:24 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* ClutterActor:scale-center-x:
|
|
|
|
*
|
|
|
|
* The horizontal center point for scaling
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("scale-center-x",
|
|
|
|
"Scale-Center-X",
|
|
|
|
"Horizontal scale center",
|
|
|
|
-G_MAXFLOAT, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
2009-01-23 11:11:24 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_SCALE_CENTER_X, pspec);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* ClutterActor:scale-center-y:
|
|
|
|
*
|
|
|
|
* The vertical center point for scaling
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("scale-center-y",
|
|
|
|
"Scale-Center-Y",
|
|
|
|
"Vertical scale center",
|
|
|
|
-G_MAXFLOAT, G_MAXFLOAT,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
2009-01-23 11:11:24 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_SCALE_CENTER_Y, pspec);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* ClutterActor:scale-gravity:
|
|
|
|
*
|
|
|
|
* The center point for scaling expressed as a #ClutterGravity
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
pspec = g_param_spec_enum ("scale-gravity",
|
|
|
|
"Scale-Gravity",
|
|
|
|
"The center of scaling",
|
|
|
|
CLUTTER_TYPE_GRAVITY,
|
|
|
|
CLUTTER_GRAVITY_NONE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
PROP_SCALE_GRAVITY,
|
|
|
|
pspec);
|
2009-01-23 11:11:24 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:rotation-angle-x:
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The rotation angle on the X axis
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_double ("rotation-angle-x",
|
|
|
|
"Rotation Angle X",
|
|
|
|
"The rotation angle on the X axis",
|
|
|
|
-G_MAXDOUBLE, G_MAXDOUBLE,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_ROTATION_ANGLE_X, pspec);
|
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:rotation-angle-y:
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The rotation angle on the Y axis
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_double ("rotation-angle-y",
|
|
|
|
"Rotation Angle Y",
|
|
|
|
"The rotation angle on the Y axis",
|
|
|
|
-G_MAXDOUBLE, G_MAXDOUBLE,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_ROTATION_ANGLE_Y, pspec);
|
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:rotation-angle-z:
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The rotation angle on the Z axis
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_double ("rotation-angle-z",
|
|
|
|
"Rotation Angle Z",
|
|
|
|
"The rotation angle on the Z axis",
|
|
|
|
-G_MAXDOUBLE, G_MAXDOUBLE,
|
|
|
|
0.0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class, PROP_ROTATION_ANGLE_Z, pspec);
|
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:rotation-center-x:
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
*
|
|
|
|
* The rotation center on the X axis.
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boxed ("rotation-center-x",
|
|
|
|
"Rotation Center X",
|
|
|
|
"The rotation center on the X axis",
|
|
|
|
CLUTTER_TYPE_VERTEX,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_ROTATION_CENTER_X,
|
|
|
|
pspec);
|
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:rotation-center-y:
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
*
|
|
|
|
* The rotation center on the Y axis.
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boxed ("rotation-center-y",
|
|
|
|
"Rotation Center Y",
|
|
|
|
"The rotation center on the Y axis",
|
|
|
|
CLUTTER_TYPE_VERTEX,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_ROTATION_CENTER_Y,
|
|
|
|
pspec);
|
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:rotation-center-z:
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
*
|
|
|
|
* The rotation center on the Z axis.
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boxed ("rotation-center-z",
|
|
|
|
"Rotation Center Z",
|
|
|
|
"The rotation center on the Z axis",
|
|
|
|
CLUTTER_TYPE_VERTEX,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_ROTATION_CENTER_Z,
|
|
|
|
pspec);
|
2009-01-23 15:55:41 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* ClutterActor:rotation-center-z-gravity:
|
|
|
|
*
|
|
|
|
* The rotation center on the Z axis expressed as a #ClutterGravity.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
pspec = g_param_spec_enum ("rotation-center-z-gravity",
|
|
|
|
"Rotation-Center-Z-Gravity",
|
|
|
|
"Center point for rotation around the Z axis",
|
|
|
|
CLUTTER_TYPE_GRAVITY,
|
|
|
|
CLUTTER_GRAVITY_NONE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_ROTATION_CENTER_Z_GRAVITY,
|
|
|
|
pspec);
|
|
|
|
|
2008-03-06 04:37:00 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:anchor-x:
|
2008-03-06 04:37:00 +00:00
|
|
|
*
|
|
|
|
* The X coordinate of an actor's anchor point, relative to
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* the actor coordinate space, in pixels
|
2008-03-06 04:37:00 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("anchor-x",
|
|
|
|
"Anchor X",
|
|
|
|
"X coordinate of the anchor point",
|
|
|
|
-G_MAXFLOAT, G_MAXFLOAT,
|
|
|
|
0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_ANCHOR_X, pspec);
|
|
|
|
|
2008-03-06 04:37:00 +00:00
|
|
|
/**
|
2008-12-12 22:47:46 +00:00
|
|
|
* ClutterActor:anchor-y:
|
2008-03-06 04:37:00 +00:00
|
|
|
*
|
|
|
|
* The Y coordinate of an actor's anchor point, relative to
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* the actor coordinate space, in pixels
|
2008-03-06 04:37:00 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
2009-05-31 14:15:46 +00:00
|
|
|
pspec = g_param_spec_float ("anchor-y",
|
|
|
|
"Anchor Y",
|
|
|
|
"Y coordinate of the anchor point",
|
|
|
|
-G_MAXFLOAT, G_MAXFLOAT,
|
|
|
|
0,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_class_install_property (object_class, PROP_ANCHOR_Y, pspec);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:anchor-gravity:
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The anchor point expressed as a #ClutterGravity
|
2009-01-22 13:07:33 +00:00
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
pspec = g_param_spec_enum ("anchor-gravity",
|
|
|
|
"Anchor-Gravity",
|
|
|
|
"The anchor point as a ClutterGravity",
|
|
|
|
CLUTTER_TYPE_GRAVITY,
|
|
|
|
CLUTTER_GRAVITY_NONE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_ANCHOR_GRAVITY, pspec);
|
|
|
|
|
2008-04-04 13:20:02 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:show-on-set-parent:
|
|
|
|
*
|
2008-04-23 13:54:13 +00:00
|
|
|
* If %TRUE, the actor is automatically shown when parented.
|
|
|
|
*
|
|
|
|
* Calling clutter_actor_hide() on an actor which has not been
|
|
|
|
* parented will set this property to %FALSE as a side effect.
|
2008-04-04 13:20:02 +00:00
|
|
|
*
|
2008-06-16 09:27:44 +00:00
|
|
|
* Since: 0.8
|
2008-04-04 13:20:02 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
pspec = g_param_spec_boolean ("show-on-set-parent",
|
|
|
|
"Show on set parent",
|
|
|
|
"Whether the actor is shown when parented",
|
|
|
|
TRUE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_SHOW_ON_SET_PARENT,
|
|
|
|
pspec);
|
2008-04-04 13:20:02 +00:00
|
|
|
|
2009-03-16 11:58:58 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor:clip-to-allocation:
|
|
|
|
*
|
|
|
|
* Whether the clip region should track the allocated area
|
|
|
|
* of the actor.
|
|
|
|
*
|
|
|
|
* This property is ignored if a clip area has been explicitly
|
|
|
|
* set using clutter_actor_set_clip().
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
pspec = g_param_spec_boolean ("clip-to-allocation",
|
|
|
|
"Clip to Allocation",
|
|
|
|
"Sets the clip region to track the "
|
|
|
|
"actor's allocation",
|
|
|
|
FALSE,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_CLIP_TO_ALLOCATION,
|
|
|
|
pspec);
|
2008-04-04 13:20:02 +00:00
|
|
|
|
2009-11-06 16:55:46 +00:00
|
|
|
pspec = g_param_spec_enum ("text-direction",
|
|
|
|
"Text Direction",
|
|
|
|
"Direction of the text",
|
|
|
|
CLUTTER_TYPE_TEXT_DIRECTION,
|
|
|
|
CLUTTER_TEXT_DIRECTION_LTR,
|
|
|
|
CLUTTER_PARAM_READWRITE);
|
|
|
|
g_object_class_install_property (object_class,
|
|
|
|
PROP_TEXT_DIRECTION,
|
|
|
|
pspec);
|
|
|
|
|
2006-07-06 18:55:46 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::destroy:
|
|
|
|
* @actor: the object which received the signal
|
|
|
|
*
|
|
|
|
* The ::destroy signal is emitted when an actor is destroyed,
|
|
|
|
* either by direct invocation of clutter_actor_destroy() or
|
|
|
|
* when the #ClutterGroup that contains the actor is destroyed.
|
|
|
|
*
|
2006-11-20 12:02:24 +00:00
|
|
|
* Since: 0.2
|
2006-07-06 18:55:46 +00:00
|
|
|
*/
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
actor_signals[DESTROY] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("destroy"),
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_CLEANUP | G_SIGNAL_NO_RECURSE | G_SIGNAL_NO_HOOKS,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, destroy),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__VOID,
|
|
|
|
G_TYPE_NONE, 0);
|
2006-07-06 18:55:46 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::show:
|
|
|
|
* @actor: the object which received the signal
|
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* The ::show signal is emitted when an actor is visible and
|
|
|
|
* rendered on the stage.
|
2007-10-12 08:17:00 +00:00
|
|
|
*
|
2006-11-20 12:02:24 +00:00
|
|
|
* Since: 0.2
|
2006-07-06 18:55:46 +00:00
|
|
|
*/
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
actor_signals[SHOW] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("show"),
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_FIRST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, show),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__VOID,
|
|
|
|
G_TYPE_NONE, 0);
|
2006-07-06 18:55:46 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::hide:
|
|
|
|
* @actor: the object which received the signal
|
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* The ::hide signal is emitted when an actor is no longer rendered
|
|
|
|
* on the stage.
|
2006-07-06 18:55:46 +00:00
|
|
|
*
|
2006-11-20 12:02:24 +00:00
|
|
|
* Since: 0.2
|
2006-07-06 18:55:46 +00:00
|
|
|
*/
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
actor_signals[HIDE] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("hide"),
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_FIRST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, hide),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__VOID,
|
2006-11-20 12:02:24 +00:00
|
|
|
G_TYPE_NONE, 0);
|
|
|
|
/**
|
|
|
|
* ClutterActor::parent-set:
|
|
|
|
* @actor: the object which received the signal
|
|
|
|
* @old_parent: the previous parent of the actor, or %NULL
|
|
|
|
*
|
|
|
|
* This signal is emitted when the parent of the actor changes.
|
|
|
|
*
|
|
|
|
* Since: 0.2
|
|
|
|
*/
|
|
|
|
actor_signals[PARENT_SET] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("parent-set"),
|
2006-11-20 12:02:24 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, parent_set),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__OBJECT,
|
|
|
|
G_TYPE_NONE, 1,
|
|
|
|
CLUTTER_TYPE_ACTOR);
|
2006-12-12 20:20:04 +00:00
|
|
|
|
2009-02-17 17:22:02 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::queue-redraw:
|
|
|
|
* @actor: the actor we're bubbling the redraw request through
|
|
|
|
* @origin: the actor which initiated the redraw request
|
|
|
|
*
|
|
|
|
* The ::queue_redraw signal is emitted when clutter_actor_queue_redraw()
|
|
|
|
* is called on @origin.
|
|
|
|
*
|
|
|
|
* The default implementation for #ClutterActor chains up to the
|
|
|
|
* parent actor and queues a redraw on the parent, thus "bubbling"
|
|
|
|
* the redraw queue up through the actor graph. The default
|
|
|
|
* implementation for #ClutterStage queues a clutter_redraw() in a
|
|
|
|
* main loop idle handler.
|
|
|
|
*
|
|
|
|
* Note that the @origin actor may be the stage, or a container; it
|
|
|
|
* does not have to be a leaf node in the actor graph.
|
|
|
|
*
|
|
|
|
* Toolkits embedding a #ClutterStage which require a redraw and
|
|
|
|
* relayout cycle can stop the emission of this signal using the
|
|
|
|
* GSignal API, redraw the UI and then call clutter_redraw()
|
|
|
|
* themselves, like:
|
|
|
|
*
|
|
|
|
* |[
|
|
|
|
* static void
|
|
|
|
* on_redraw_complete (void)
|
|
|
|
* {
|
|
|
|
* /* execute the Clutter drawing pipeline */
|
|
|
|
* clutter_redraw ();
|
|
|
|
* }
|
|
|
|
*
|
|
|
|
* static void
|
|
|
|
* on_stage_queue_redraw (ClutterStage *stage)
|
|
|
|
* {
|
|
|
|
* /* this prevents the default handler to run */
|
|
|
|
* g_signal_stop_emission_by_name (stage, "queue-redraw");
|
|
|
|
*
|
|
|
|
* /* queue a redraw with the host toolkit and call
|
|
|
|
* * a function when the redraw has been completed
|
|
|
|
* */
|
|
|
|
* queue_a_redraw (G_CALLBACK (on_redraw_complete));
|
|
|
|
* }
|
|
|
|
* ]|
|
|
|
|
*
|
|
|
|
* <note><para>This signal is emitted before the Clutter paint
|
|
|
|
* pipeline is executed. If you want to know when the pipeline has
|
|
|
|
* been completed you should connect to the ::paint signal on the
|
|
|
|
* Stage with g_signal_connect_after().</para></note>
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
actor_signals[QUEUE_REDRAW] =
|
|
|
|
g_signal_new (I_("queue-redraw"),
|
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, queue_redraw),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__OBJECT,
|
|
|
|
G_TYPE_NONE, 1,
|
|
|
|
CLUTTER_TYPE_ACTOR);
|
|
|
|
|
2009-08-08 01:16:04 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::queue-relayout
|
|
|
|
* @actor: the actor being queued for relayout
|
|
|
|
*
|
|
|
|
* The ::queue_layout signal is emitted when clutter_actor_queue_relayout()
|
|
|
|
* is called on an actor.
|
|
|
|
*
|
|
|
|
* The default implementation for #ClutterActor chains up to the
|
|
|
|
* parent actor and queues a relayout on the parent, thus "bubbling"
|
|
|
|
* the relayout queue up through the actor graph.
|
|
|
|
*
|
|
|
|
* The main purpose of this signal is to allow relayout to be propagated
|
|
|
|
* properly in the procense of #ClutterClone actors. Applications will
|
|
|
|
* not normally need to connect to this signal.
|
|
|
|
*
|
|
|
|
* Since: 1.2
|
|
|
|
*/
|
|
|
|
actor_signals[QUEUE_RELAYOUT] =
|
|
|
|
g_signal_new (I_("queue-relayout"),
|
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, queue_relayout),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__VOID,
|
|
|
|
G_TYPE_NONE, 0);
|
|
|
|
|
2007-08-13 20:48:01 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::event:
|
|
|
|
* @actor: the actor which received the event
|
|
|
|
* @event: a #ClutterEvent
|
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* The ::event signal is emitted each time an event is received
|
2007-12-17 15:47:19 +00:00
|
|
|
* by the @actor. This signal will be emitted on every actor,
|
|
|
|
* following the hierarchy chain, until it reaches the top-level
|
|
|
|
* container (the #ClutterStage).
|
2007-08-13 20:48:01 +00:00
|
|
|
*
|
2007-10-02 14:03:36 +00:00
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
|
|
|
*
|
2007-08-13 20:48:01 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
actor_signals[EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("event"),
|
2007-08-13 20:48:01 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, event),
|
2007-10-02 14:03:36 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
2007-08-13 20:48:01 +00:00
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
|
|
|
/**
|
|
|
|
* ClutterActor::button-press-event:
|
|
|
|
* @actor: the actor which received the event
|
|
|
|
* @event: a #ClutterButtonEvent
|
|
|
|
*
|
|
|
|
* The ::button-press-event signal is emitted each time a mouse button
|
|
|
|
* is pressed on @actor.
|
|
|
|
*
|
2007-10-02 14:03:36 +00:00
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
|
|
|
*
|
2007-08-13 20:48:01 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
actor_signals[BUTTON_PRESS_EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("button-press-event"),
|
2007-08-13 20:48:01 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, button_press_event),
|
2007-10-02 14:03:36 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
2007-08-13 20:48:01 +00:00
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
|
|
|
/**
|
|
|
|
* ClutterActor::button-release-event:
|
|
|
|
* @actor: the actor which received the event
|
|
|
|
* @event: a #ClutterButtonEvent
|
|
|
|
*
|
|
|
|
* The ::button-release-event signal is emitted each time a mouse button
|
|
|
|
* is released on @actor.
|
|
|
|
*
|
2007-10-02 14:03:36 +00:00
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
|
|
|
*
|
2007-08-13 20:48:01 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
actor_signals[BUTTON_RELEASE_EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("button-release-event"),
|
2007-08-13 20:48:01 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, button_release_event),
|
2007-10-02 14:03:36 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
2007-08-13 20:48:01 +00:00
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
|
|
|
/**
|
|
|
|
* ClutterActor::scroll-event:
|
|
|
|
* @actor: the actor which received the event
|
|
|
|
* @event: a #ClutterScrollEvent
|
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* The ::scroll-event signal is emitted each time the mouse is
|
2007-08-13 20:48:01 +00:00
|
|
|
* scrolled on @actor
|
|
|
|
*
|
2007-10-02 14:03:36 +00:00
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
2007-08-13 20:48:01 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
actor_signals[SCROLL_EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("scroll-event"),
|
2007-08-13 20:48:01 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, scroll_event),
|
2007-10-02 14:03:36 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
2007-08-13 20:48:01 +00:00
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
|
|
|
/**
|
|
|
|
* ClutterActor::key-press-event:
|
|
|
|
* @actor: the actor which received the event
|
|
|
|
* @event: a #ClutterKeyEvent
|
|
|
|
*
|
|
|
|
* The ::key-press-event signal is emitted each time a keyboard button
|
2008-02-15 14:39:25 +00:00
|
|
|
* is pressed while @actor has key focus (see clutter_stage_set_key_focus()).
|
2007-08-13 20:48:01 +00:00
|
|
|
*
|
2007-10-02 14:03:36 +00:00
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
|
|
|
*
|
2007-08-13 20:48:01 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
actor_signals[KEY_PRESS_EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("key-press-event"),
|
2007-08-13 20:48:01 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, key_press_event),
|
2007-10-02 14:03:36 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
2007-08-13 20:48:01 +00:00
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
|
|
|
/**
|
|
|
|
* ClutterActor::key-release-event:
|
|
|
|
* @actor: the actor which received the event
|
|
|
|
* @event: a #ClutterKeyEvent
|
|
|
|
*
|
|
|
|
* The ::key-release-event signal is emitted each time a keyboard button
|
2008-02-15 14:39:25 +00:00
|
|
|
* is released while @actor has key focus (see
|
|
|
|
* clutter_stage_set_key_focus()).
|
2007-08-13 20:48:01 +00:00
|
|
|
*
|
2007-10-02 14:03:36 +00:00
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
|
|
|
*
|
2007-08-13 20:48:01 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
actor_signals[KEY_RELEASE_EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("key-release-event"),
|
2007-08-13 20:48:01 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, key_release_event),
|
2007-10-02 14:03:36 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
2007-08-13 20:48:01 +00:00
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
|
|
|
/**
|
|
|
|
* ClutterActor::motion-event:
|
|
|
|
* @actor: the actor which received the event
|
|
|
|
* @event: a #ClutterMotionEvent
|
|
|
|
*
|
|
|
|
* The ::motion-event signal is emitted each time the mouse pointer is
|
2008-02-15 14:39:25 +00:00
|
|
|
* moved over @actor.
|
2007-08-13 20:48:01 +00:00
|
|
|
*
|
2007-10-02 14:03:36 +00:00
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
|
|
|
*
|
2007-08-13 20:48:01 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
actor_signals[MOTION_EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("motion-event"),
|
2007-08-13 20:48:01 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, motion_event),
|
2007-10-02 14:03:36 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
2007-08-13 20:48:01 +00:00
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
|
|
|
|
|
|
|
/**
|
2009-05-06 16:56:40 +00:00
|
|
|
* ClutterActor::key-focus-in:
|
2007-08-13 20:48:01 +00:00
|
|
|
* @actor: the actor which now has key focus
|
|
|
|
*
|
|
|
|
* The ::focus-in signal is emitted when @actor recieves key focus.
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
2009-05-06 16:56:40 +00:00
|
|
|
actor_signals[KEY_FOCUS_IN] =
|
|
|
|
g_signal_new (I_("key-focus-in"),
|
2007-08-13 20:48:01 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
2009-05-06 16:56:40 +00:00
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, key_focus_in),
|
2007-08-13 20:48:01 +00:00
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__VOID,
|
|
|
|
G_TYPE_NONE, 0);
|
|
|
|
|
|
|
|
/**
|
2009-05-06 16:56:40 +00:00
|
|
|
* ClutterActor::key-focus-out:
|
2007-08-13 20:48:01 +00:00
|
|
|
* @actor: the actor which now has key focus
|
|
|
|
*
|
2009-05-06 16:56:40 +00:00
|
|
|
* The ::key-focus-out signal is emitted when @actor loses key focus.
|
2007-08-13 20:48:01 +00:00
|
|
|
*
|
2007-10-09 23:45:49 +00:00
|
|
|
* Since: 0.6
|
2007-08-13 20:48:01 +00:00
|
|
|
*/
|
2009-05-06 16:56:40 +00:00
|
|
|
actor_signals[KEY_FOCUS_OUT] =
|
|
|
|
g_signal_new (I_("key-focus-out"),
|
2007-08-13 20:48:01 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
2009-05-06 16:56:40 +00:00
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, key_focus_out),
|
2007-08-13 20:48:01 +00:00
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__VOID,
|
|
|
|
G_TYPE_NONE, 0);
|
|
|
|
|
2007-10-09 23:45:49 +00:00
|
|
|
/**
|
2007-10-10 14:29:29 +00:00
|
|
|
* ClutterActor::enter-event:
|
2007-10-09 23:45:49 +00:00
|
|
|
* @actor: the actor which the pointer has entered.
|
2007-12-17 15:47:19 +00:00
|
|
|
* @event: a #ClutterCrossingEvent
|
2007-10-09 23:45:49 +00:00
|
|
|
*
|
2007-10-12 08:17:00 +00:00
|
|
|
* The ::enter-event signal is emitted when the pointer enters the @actor
|
2007-10-09 23:45:49 +00:00
|
|
|
*
|
2007-12-17 15:47:19 +00:00
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
|
|
|
*
|
2007-10-09 23:45:49 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
actor_signals[ENTER_EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("enter-event"),
|
2007-10-09 23:45:49 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
2007-11-07 16:40:30 +00:00
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, enter_event),
|
2007-10-10 13:04:34 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
2007-10-09 23:45:49 +00:00
|
|
|
|
|
|
|
/**
|
2007-10-10 14:29:29 +00:00
|
|
|
* ClutterActor::leave-event:
|
2007-10-09 23:45:49 +00:00
|
|
|
* @actor: the actor which the pointer has left
|
2007-12-17 15:47:19 +00:00
|
|
|
* @event: a #ClutterCrossingEvent
|
2007-10-09 23:45:49 +00:00
|
|
|
*
|
2007-10-10 14:29:29 +00:00
|
|
|
* The ::leave-event signal is emitted when the pointer leaves the @actor.
|
2007-10-09 23:45:49 +00:00
|
|
|
*
|
2007-12-17 15:47:19 +00:00
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
|
|
|
*
|
2007-10-09 23:45:49 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
actor_signals[LEAVE_EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("leave-event"),
|
2007-10-09 23:45:49 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
2007-11-07 16:40:30 +00:00
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, leave_event),
|
2007-10-10 13:04:34 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
2007-10-09 23:45:49 +00:00
|
|
|
|
2007-10-10 13:04:34 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::captured-event:
|
2007-12-17 15:47:19 +00:00
|
|
|
* @actor: the actor which received the signal
|
|
|
|
* @event: a #ClutterEvent
|
2007-10-10 13:04:34 +00:00
|
|
|
*
|
2007-10-10 14:29:29 +00:00
|
|
|
* The ::captured-event signal is emitted when an event is captured
|
2007-12-17 15:47:19 +00:00
|
|
|
* by Clutter. This signal will be emitted starting from the top-level
|
|
|
|
* container (the #ClutterStage) to the actor which received the event
|
|
|
|
* going down the hierarchy. This signal can be used to intercept every
|
|
|
|
* event before the specialized events (like
|
|
|
|
* ClutterActor::button-press-event or ::key-released-event) are
|
|
|
|
* emitted.
|
|
|
|
*
|
|
|
|
* Return value: %TRUE if the event has been handled by the actor,
|
|
|
|
* or %FALSE to continue the emission.
|
2007-10-10 13:04:34 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
2007-11-07 16:40:30 +00:00
|
|
|
actor_signals[CAPTURED_EVENT] =
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_signal_new (I_("captured-event"),
|
2007-10-10 13:04:34 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
2007-11-07 16:40:30 +00:00
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, captured_event),
|
2007-10-10 13:04:34 +00:00
|
|
|
_clutter_boolean_handled_accumulator, NULL,
|
|
|
|
clutter_marshal_BOOLEAN__BOXED,
|
|
|
|
G_TYPE_BOOLEAN, 1,
|
|
|
|
CLUTTER_TYPE_EVENT | G_SIGNAL_TYPE_STATIC_SCOPE);
|
2007-08-13 20:48:01 +00:00
|
|
|
|
2008-06-05 14:32:18 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::paint:
|
|
|
|
* @actor: the #ClutterActor that received the signal
|
|
|
|
*
|
|
|
|
* The ::paint signal is emitted each time an actor is being painted.
|
|
|
|
*
|
|
|
|
* Subclasses of #ClutterActor should override the class signal handler
|
|
|
|
* and paint themselves in that function.
|
|
|
|
*
|
|
|
|
* It is possible to connect a handler to the ::paint signal in order
|
|
|
|
* to set up some custom aspect of a paint.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
actor_signals[PAINT] =
|
|
|
|
g_signal_new (I_("paint"),
|
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, paint),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__VOID,
|
|
|
|
G_TYPE_NONE, 0);
|
|
|
|
/**
|
|
|
|
* ClutterActor::realize:
|
|
|
|
* @actor: the #ClutterActor that received the signal
|
|
|
|
*
|
|
|
|
* The ::realize signal is emitted each time an actor is being
|
|
|
|
* realized.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
actor_signals[REALIZE] =
|
|
|
|
g_signal_new (I_("realize"),
|
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, realize),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__VOID,
|
|
|
|
G_TYPE_NONE, 0);
|
|
|
|
/**
|
|
|
|
* ClutterActor::unrealize:
|
|
|
|
* @actor: the #ClutterActor that received the signal
|
|
|
|
*
|
|
|
|
* The ::unrealize signal is emitted each time an actor is being
|
|
|
|
* unrealized.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
actor_signals[UNREALIZE] =
|
2008-06-12 17:24:07 +00:00
|
|
|
g_signal_new (I_("unrealize"),
|
2008-06-05 14:32:18 +00:00
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, unrealize),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__VOID,
|
|
|
|
G_TYPE_NONE, 0);
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::map:
|
|
|
|
* @actor: the #ClutterActor to map
|
|
|
|
*
|
|
|
|
* The ::map virtual functon must be overridden in order to call
|
|
|
|
* clutter_actor_map() on any child actors if the actor is not a
|
|
|
|
* #ClutterContainer. When overriding, it is mandatory to chain up
|
|
|
|
* to the parent implementation.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
|
|
|
* ClutterActor::unmap:
|
|
|
|
* @actor: the #ClutterActor to unmap
|
|
|
|
*
|
|
|
|
* The ::unmap virtual functon must be overridden in order to call
|
|
|
|
* clutter_actor_unmap() on any child actors if the actor is not a
|
|
|
|
* #ClutterContainer. When overriding, it is mandatory to chain up
|
|
|
|
* to the parent implementation.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
|
2009-01-23 11:34:13 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::pick:
|
|
|
|
* @actor: the #ClutterActor that received the signal
|
|
|
|
* @color: the #ClutterColor to be used when picking
|
|
|
|
*
|
|
|
|
* The ::pick signal is emitted each time an actor is being painted
|
|
|
|
* in "pick mode". The pick mode is used to identify the actor during
|
|
|
|
* the event handling phase, or by clutter_stage_get_actor_at_pos().
|
|
|
|
* The actor should paint its shape using the passed @pick_color.
|
|
|
|
*
|
|
|
|
* Subclasses of #ClutterActor should override the class signal handler
|
|
|
|
* and paint themselves in that function.
|
|
|
|
*
|
|
|
|
* It is possible to connect a handler to the ::pick signal in order
|
|
|
|
* to set up some custom aspect of a paint in pick mode.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
actor_signals[PICK] =
|
|
|
|
g_signal_new (I_("pick"),
|
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
G_STRUCT_OFFSET (ClutterActorClass, pick),
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__BOXED,
|
|
|
|
G_TYPE_NONE, 1,
|
|
|
|
CLUTTER_TYPE_COLOR);
|
|
|
|
|
2009-06-04 11:00:58 +00:00
|
|
|
/**
|
|
|
|
* ClutterActor::allocation-changed:
|
|
|
|
* @actor: the #ClutterActor that emitted the signal
|
|
|
|
* @box: a #ClutterActorBox with the new allocation
|
|
|
|
* @flags: #ClutterAllocationFlags for the allocation
|
|
|
|
*
|
|
|
|
* The ::allocation-changed signal is emitted when the
|
|
|
|
* #ClutterActor:allocation property changes. Usually, application
|
|
|
|
* code should just use the notifications for the :allocation property
|
|
|
|
* but if you want to track the allocation flags as well, for instance
|
|
|
|
* to know whether the absolute origin of @actor changed, then you might
|
|
|
|
* want use this signal instead.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
actor_signals[ALLOCATION_CHANGED] =
|
|
|
|
g_signal_new (I_("allocation-changed"),
|
|
|
|
G_TYPE_FROM_CLASS (object_class),
|
|
|
|
G_SIGNAL_RUN_LAST,
|
|
|
|
0,
|
|
|
|
NULL, NULL,
|
|
|
|
clutter_marshal_VOID__BOXED_FLAGS,
|
|
|
|
G_TYPE_NONE, 2,
|
|
|
|
CLUTTER_TYPE_ACTOR_BOX,
|
|
|
|
CLUTTER_TYPE_ALLOCATION_FLAGS);
|
|
|
|
|
2006-12-12 20:20:04 +00:00
|
|
|
klass->show = clutter_actor_real_show;
|
|
|
|
klass->show_all = clutter_actor_show;
|
|
|
|
klass->hide = clutter_actor_real_hide;
|
|
|
|
klass->hide_all = clutter_actor_hide;
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
klass->map = clutter_actor_real_map;
|
|
|
|
klass->unmap = clutter_actor_real_unmap;
|
|
|
|
klass->unrealize = clutter_actor_real_unrealize;
|
2007-11-18 23:24:55 +00:00
|
|
|
klass->pick = clutter_actor_real_pick;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
klass->get_preferred_width = clutter_actor_real_get_preferred_width;
|
|
|
|
klass->get_preferred_height = clutter_actor_real_get_preferred_height;
|
|
|
|
klass->allocate = clutter_actor_real_allocate;
|
2009-02-17 17:22:02 +00:00
|
|
|
klass->queue_redraw = clutter_actor_real_queue_redraw;
|
2009-08-08 01:16:04 +00:00
|
|
|
klass->queue_relayout = clutter_actor_real_queue_relayout;
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
klass->apply_transform = clutter_actor_real_apply_transform;
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_init (ClutterActor *self)
|
|
|
|
{
|
2007-10-02 14:45:40 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
2007-10-12 08:17:00 +00:00
|
|
|
self->priv = priv = CLUTTER_ACTOR_GET_PRIVATE (self);
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-10-02 14:45:40 +00:00
|
|
|
priv->parent_actor = NULL;
|
|
|
|
priv->has_clip = FALSE;
|
|
|
|
priv->opacity = 0xff;
|
2008-03-07 17:34:40 +00:00
|
|
|
priv->id = clutter_id_pool_add (CLUTTER_CONTEXT()->id_pool, self);
|
2009-01-20 16:20:54 +00:00
|
|
|
priv->scale_x = 1.0;
|
|
|
|
priv->scale_y = 1.0;
|
2008-02-29 14:54:28 +00:00
|
|
|
priv->shader_data = NULL;
|
2008-04-04 13:20:02 +00:00
|
|
|
priv->show_on_set_parent = TRUE;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->needs_width_request = TRUE;
|
|
|
|
priv->needs_height_request = TRUE;
|
|
|
|
priv->needs_allocation = TRUE;
|
2008-01-31 14:21:36 +00:00
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
priv->cached_width_age = 1;
|
|
|
|
priv->cached_height_age = 1;
|
|
|
|
|
2009-01-17 16:51:03 +00:00
|
|
|
priv->opacity_parent = NULL;
|
|
|
|
priv->enable_model_view_transform = TRUE;
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
memset (priv->clip, 0, sizeof (gfloat) * 4);
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_destroy:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Destroys an actor. When an actor is destroyed, it will break any
|
|
|
|
* references it holds to other objects. If the actor is inside a
|
2007-08-13 17:01:07 +00:00
|
|
|
* container, the actor will be removed.
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* When you destroy a container, its children will be destroyed as well.
|
2007-11-23 11:20:14 +00:00
|
|
|
*
|
|
|
|
* Note: you cannot destroy the #ClutterStage returned by
|
|
|
|
* clutter_stage_get_default().
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_destroy (ClutterActor *self)
|
|
|
|
{
|
2007-11-23 11:20:14 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2007-11-23 11:20:14 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
2008-06-03 20:15:11 +00:00
|
|
|
g_object_ref (self);
|
|
|
|
|
2008-06-13 13:59:07 +00:00
|
|
|
/* avoid recursion while destroying */
|
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_DESTRUCTION))
|
2007-10-17 16:43:46 +00:00
|
|
|
{
|
2008-06-13 13:59:07 +00:00
|
|
|
CLUTTER_SET_PRIVATE_FLAGS (self, CLUTTER_ACTOR_IN_DESTRUCTION);
|
2007-10-17 16:43:46 +00:00
|
|
|
|
2009-06-03 11:59:16 +00:00
|
|
|
/* if we are destroying we want to unrealize ourselves
|
|
|
|
* first before the dispose run removes the parent
|
|
|
|
*/
|
2009-06-04 15:50:19 +00:00
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL))
|
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_MAKE_UNREALIZED);
|
2009-06-03 11:59:16 +00:00
|
|
|
|
2008-06-13 13:59:07 +00:00
|
|
|
g_object_run_dispose (G_OBJECT (self));
|
2007-10-17 16:43:46 +00:00
|
|
|
|
2009-10-09 16:18:34 +00:00
|
|
|
CLUTTER_UNSET_PRIVATE_FLAGS (self, CLUTTER_ACTOR_IN_DESTRUCTION);
|
2008-06-13 13:59:07 +00:00
|
|
|
}
|
2007-11-27 11:05:26 +00:00
|
|
|
|
|
|
|
g_object_unref (self);
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_queue_redraw:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2007-10-12 08:17:00 +00:00
|
|
|
* Queues up a redraw of an actor and any children. The redraw occurs
|
|
|
|
* once the main loop becomes idle (after the current batch of events
|
2006-06-13 13:17:45 +00:00
|
|
|
* has been processed, roughly).
|
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* Applications rarely need to call this, as redraws are handled
|
|
|
|
* automatically by modification functions.
|
2009-01-22 13:58:50 +00:00
|
|
|
*
|
|
|
|
* This function will not do anything if @self is not visible, or
|
|
|
|
* if the actor is inside an invisible part of the scenegraph.
|
2009-05-20 16:55:29 +00:00
|
|
|
*
|
|
|
|
* Also be aware that painting is a NOP for actors with an opacity of
|
|
|
|
* 0
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_queue_redraw (ClutterActor *self)
|
|
|
|
{
|
2008-03-28 22:50:55 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2007-08-08 10:20:14 +00:00
|
|
|
|
2009-02-17 17:22:02 +00:00
|
|
|
clutter_actor_queue_redraw_with_origin (self, self);
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* clutter_actor_queue_relayout:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Indicates that the actor's size request or other layout-affecting
|
|
|
|
* properties may have changed. This function is used inside #ClutterActor
|
|
|
|
* subclass implementations, not by applications directly.
|
|
|
|
*
|
|
|
|
* Queueing a new layout automatically queues a redraw as well.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_queue_relayout (ClutterActor *self)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
2009-08-05 16:46:18 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->needs_width_request &&
|
|
|
|
priv->needs_height_request &&
|
|
|
|
priv->needs_allocation)
|
|
|
|
return; /* save some cpu cycles */
|
|
|
|
|
2009-10-23 09:17:40 +00:00
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL) &&
|
|
|
|
(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_RELAYOUT))
|
|
|
|
{
|
|
|
|
g_warning ("The actor '%s' is currently inside an allocation "
|
|
|
|
"cycle; calling clutter_actor_queue_relayout() is "
|
|
|
|
"not allowed",
|
|
|
|
priv->name ? priv->name
|
|
|
|
: G_OBJECT_TYPE_NAME (self));
|
|
|
|
}
|
|
|
|
|
2009-08-08 01:16:04 +00:00
|
|
|
g_signal_emit (self, actor_signals[QUEUE_RELAYOUT], 0);
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* clutter_actor_get_preferred_size:
|
|
|
|
* @self: a #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @min_width_p: (out) (allow-none): return location for the minimum
|
|
|
|
* width, or %NULL
|
|
|
|
* @min_height_p: (out) (allow-none): return location for the minimum
|
|
|
|
* height, or %NULL
|
|
|
|
* @natural_width_p: (out) (allow-none): return location for the natural
|
|
|
|
* width, or %NULL
|
|
|
|
* @natural_height_p: (out) (allow-none): return location for the natural
|
|
|
|
* height, or %NULL
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Computes the preferred minimum and natural size of an actor, taking into
|
|
|
|
* account the actor's geometry management (either height-for-width
|
|
|
|
* or width-for-height).
|
|
|
|
*
|
|
|
|
* The width and height used to compute the preferred height and preferred
|
|
|
|
* width are the actor's natural ones.
|
|
|
|
*
|
|
|
|
* If you need to control the height for the preferred width, or the width for
|
|
|
|
* the preferred height, you should use clutter_actor_get_preferred_width()
|
|
|
|
* and clutter_actor_get_preferred_height(), and check the actor's preferred
|
|
|
|
* geometry management using the #ClutterActor:request-mode property.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_preferred_size (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat *min_width_p,
|
|
|
|
gfloat *min_height_p,
|
|
|
|
gfloat *natural_width_p,
|
|
|
|
gfloat *natural_height_p)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat for_width, for_height;
|
|
|
|
gfloat min_width, min_height;
|
|
|
|
gfloat natural_width, natural_height;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
for_width = for_height = 0;
|
|
|
|
min_width = min_height = 0;
|
|
|
|
natural_width = natural_height = 0;
|
|
|
|
|
|
|
|
if (priv->request_mode == CLUTTER_REQUEST_HEIGHT_FOR_WIDTH)
|
|
|
|
{
|
2009-04-15 16:25:50 +00:00
|
|
|
CLUTTER_NOTE (LAYOUT, "Preferred size (height-for-width)");
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_get_preferred_width (self, -1,
|
|
|
|
&min_width,
|
|
|
|
&natural_width);
|
|
|
|
clutter_actor_get_preferred_height (self, natural_width,
|
|
|
|
&min_height,
|
|
|
|
&natural_height);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2009-04-15 16:25:50 +00:00
|
|
|
CLUTTER_NOTE (LAYOUT, "Preferred size (width-for-height)");
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_get_preferred_height (self, -1,
|
|
|
|
&min_height,
|
|
|
|
&natural_height);
|
|
|
|
clutter_actor_get_preferred_width (self, natural_height,
|
|
|
|
&min_width,
|
|
|
|
&natural_width);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (min_width_p)
|
|
|
|
*min_width_p = min_width;
|
|
|
|
|
|
|
|
if (min_height_p)
|
|
|
|
*min_height_p = min_height;
|
|
|
|
|
|
|
|
if (natural_width_p)
|
|
|
|
*natural_width_p = natural_width;
|
|
|
|
|
|
|
|
if (natural_height_p)
|
|
|
|
*natural_height_p = natural_height;
|
|
|
|
}
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
/* looks for a cached size request for this for_size. If not
|
|
|
|
* found, returns the oldest entry so it can be overwritten */
|
|
|
|
static gboolean
|
|
|
|
_clutter_actor_get_cached_size_request (gfloat for_size,
|
|
|
|
SizeRequest *cached_size_requests,
|
|
|
|
SizeRequest **result)
|
|
|
|
{
|
|
|
|
gboolean found_free_cache;
|
|
|
|
guint i;
|
|
|
|
|
|
|
|
found_free_cache = FALSE;
|
|
|
|
*result = &cached_size_requests[0];
|
|
|
|
|
|
|
|
for (i = 0; i < N_CACHED_SIZE_REQUESTS; i++)
|
|
|
|
{
|
|
|
|
SizeRequest *sr;
|
|
|
|
|
|
|
|
sr = &cached_size_requests[i];
|
|
|
|
|
|
|
|
if (sr->age > 0 &&
|
|
|
|
sr->for_size == for_size)
|
|
|
|
{
|
2009-12-04 17:38:26 +00:00
|
|
|
CLUTTER_NOTE (LAYOUT, "Size cache hit for size: %.2f", for_size);
|
2009-11-12 01:40:57 +00:00
|
|
|
*result = sr;
|
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
else if (sr->age < (*result)->age)
|
|
|
|
{
|
|
|
|
*result = sr;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-12-04 17:38:26 +00:00
|
|
|
CLUTTER_NOTE (LAYOUT, "Size cache miss for size: %.2f", for_size);
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_preferred_width:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* @for_height: available height when computing the preferred width,
|
|
|
|
* or a negative value to indicate that no height is defined
|
2009-07-02 13:21:16 +00:00
|
|
|
* @min_width_p: (out) (allow-none): return location for minimum width,
|
|
|
|
* or %NULL
|
|
|
|
* @natural_width_p: (out) (allow-none): return location for the natural
|
|
|
|
* width, or %NULL
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Computes the requested minimum and natural widths for an actor,
|
|
|
|
* optionally depending on the specified height, or if they are
|
|
|
|
* already computed, returns the cached values.
|
|
|
|
*
|
|
|
|
* An actor may not get its request - depending on the layout
|
|
|
|
* manager that's in effect.
|
|
|
|
*
|
|
|
|
* A request should not incorporate the actor's scale or anchor point;
|
|
|
|
* those transformations do not affect layout, only rendering.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_get_preferred_width (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat for_height,
|
|
|
|
gfloat *min_width_p,
|
|
|
|
gfloat *natural_width_p)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
|
|
|
ClutterActorClass *klass;
|
|
|
|
ClutterActorPrivate *priv;
|
2009-11-12 01:40:57 +00:00
|
|
|
gboolean found_in_cache;
|
|
|
|
SizeRequest *cached_size_request;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
klass = CLUTTER_ACTOR_GET_CLASS (self);
|
|
|
|
priv = self->priv;
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
found_in_cache = FALSE;
|
|
|
|
cached_size_request = &priv->width_requests[0];
|
|
|
|
|
|
|
|
if (!priv->needs_width_request)
|
|
|
|
found_in_cache = _clutter_actor_get_cached_size_request (for_height,
|
|
|
|
priv->width_requests,
|
|
|
|
&cached_size_request);
|
|
|
|
|
|
|
|
if (!found_in_cache)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat min_width, natural_width;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
min_width = natural_width = 0;
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
CLUTTER_NOTE (LAYOUT, "Width request for %.2f px", for_height);
|
2009-04-15 16:25:50 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
klass->get_preferred_width (self, for_height,
|
|
|
|
&min_width,
|
|
|
|
&natural_width);
|
|
|
|
|
2009-06-06 15:37:18 +00:00
|
|
|
/* Due to accumulated float errors, it's better not to warn
|
|
|
|
* on this, but just fix it.
|
|
|
|
*/
|
|
|
|
if (natural_width < min_width)
|
|
|
|
natural_width = min_width;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
cached_size_request->min_size = min_width;
|
|
|
|
cached_size_request->natural_size = natural_width;
|
|
|
|
cached_size_request->for_size = for_height;
|
|
|
|
cached_size_request->age = priv->cached_width_age;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
priv->cached_width_age ++;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->needs_width_request = FALSE;
|
|
|
|
}
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
if (!priv->min_width_set)
|
|
|
|
priv->request_min_width = cached_size_request->min_size;
|
|
|
|
|
|
|
|
if (!priv->natural_width_set)
|
|
|
|
priv->request_natural_width = cached_size_request->natural_size;
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (min_width_p)
|
|
|
|
*min_width_p = priv->request_min_width;
|
|
|
|
|
|
|
|
if (natural_width_p)
|
|
|
|
*natural_width_p = priv->request_natural_width;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_preferred_height:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
* @for_width: available width to assume in computing desired height,
|
|
|
|
* or a negative value to indicate that no width is defined
|
2009-07-02 13:21:16 +00:00
|
|
|
* @min_height_p: (out) (allow-none): return location for minimum height,
|
|
|
|
* or %NULL
|
|
|
|
* @natural_height_p: (out) (allow-none): return location for natural
|
|
|
|
* height, or %NULL
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Computes the requested minimum and natural heights for an actor,
|
|
|
|
* or if they are already computed, returns the cached values.
|
|
|
|
*
|
|
|
|
* An actor may not get its request - depending on the layout
|
|
|
|
* manager that's in effect.
|
|
|
|
*
|
|
|
|
* A request should not incorporate the actor's scale or anchor point;
|
|
|
|
* those transformations do not affect layout, only rendering.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_preferred_height (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat for_width,
|
|
|
|
gfloat *min_height_p,
|
|
|
|
gfloat *natural_height_p)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
|
|
|
ClutterActorClass *klass;
|
|
|
|
ClutterActorPrivate *priv;
|
2009-11-12 01:40:57 +00:00
|
|
|
gboolean found_in_cache;
|
|
|
|
SizeRequest *cached_size_request;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
klass = CLUTTER_ACTOR_GET_CLASS (self);
|
|
|
|
priv = self->priv;
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
found_in_cache = FALSE;
|
|
|
|
cached_size_request = &priv->height_requests[0];
|
|
|
|
|
|
|
|
if (!priv->needs_height_request)
|
|
|
|
found_in_cache = _clutter_actor_get_cached_size_request (for_width,
|
|
|
|
priv->height_requests,
|
|
|
|
&cached_size_request);
|
|
|
|
|
|
|
|
if (!found_in_cache)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat min_height, natural_height;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
min_height = natural_height = 0;
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
CLUTTER_NOTE (LAYOUT, "Height request for %.2f px", for_width);
|
2009-04-15 16:25:50 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
klass->get_preferred_height (self, for_width,
|
|
|
|
&min_height,
|
|
|
|
&natural_height);
|
|
|
|
|
2009-06-06 15:37:18 +00:00
|
|
|
/* Due to accumulated float errors, it's better not to warn
|
|
|
|
* on this, but just fix it.
|
|
|
|
*/
|
|
|
|
if (natural_height < min_height)
|
|
|
|
natural_height = min_height;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
if (!priv->min_height_set)
|
2009-11-12 01:40:57 +00:00
|
|
|
{
|
|
|
|
priv->request_min_height = min_height;
|
|
|
|
}
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
if (!priv->natural_height_set)
|
2009-11-12 01:40:57 +00:00
|
|
|
{
|
|
|
|
priv->request_natural_height = natural_height;
|
|
|
|
}
|
|
|
|
|
|
|
|
cached_size_request->min_size = min_height;
|
|
|
|
cached_size_request->natural_size = natural_height;
|
|
|
|
cached_size_request->for_size = for_width;
|
|
|
|
cached_size_request->age = priv->cached_height_age;
|
|
|
|
|
|
|
|
priv->cached_height_age ++;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
priv->needs_height_request = FALSE;
|
|
|
|
}
|
|
|
|
|
2009-11-12 01:40:57 +00:00
|
|
|
if (!priv->min_height_set)
|
|
|
|
priv->request_min_height = cached_size_request->min_size;
|
|
|
|
|
|
|
|
if (!priv->natural_height_set)
|
|
|
|
priv->request_natural_height = cached_size_request->natural_size;
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (min_height_p)
|
|
|
|
*min_height_p = priv->request_min_height;
|
|
|
|
|
|
|
|
if (natural_height_p)
|
|
|
|
*natural_height_p = priv->request_natural_height;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_allocation_box:
|
|
|
|
* @self: A #ClutterActor
|
2009-02-26 19:42:19 +00:00
|
|
|
* @box: (out): the function fills this in with the actor's allocation
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Gets the layout box an actor has been assigned. The allocation can
|
|
|
|
* only be assumed valid inside a paint() method; anywhere else, it
|
|
|
|
* may be out-of-date.
|
|
|
|
*
|
|
|
|
* An allocation does not incorporate the actor's scale or anchor point;
|
|
|
|
* those transformations do not affect layout, only rendering.
|
|
|
|
*
|
|
|
|
* <note>Do not call any of the clutter_actor_get_allocation_*() family
|
|
|
|
* of functions inside the implementation of the get_preferred_width()
|
|
|
|
* or get_preferred_height() virtual functions.</note>
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_allocation_box (ClutterActor *self,
|
|
|
|
ClutterActorBox *box)
|
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2009-04-15 16:25:50 +00:00
|
|
|
/* XXX - if needs_allocation=TRUE, we can either 1) g_return_if_fail,
|
|
|
|
* which limits calling get_allocation to inside paint() basically; or
|
|
|
|
* we can 2) force a layout, which could be expensive if someone calls
|
|
|
|
* get_allocation somewhere silly; or we can 3) just return the latest
|
|
|
|
* value, allowing it to be out-of-date, and assume people know what
|
|
|
|
* they are doing.
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* The least-surprises approach that keeps existing code working is
|
|
|
|
* likely to be 2). People can end up doing some inefficient things,
|
|
|
|
* though, and in general code that requires 2) is probably broken.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* this implements 2) */
|
|
|
|
if (G_UNLIKELY (self->priv->needs_allocation))
|
|
|
|
{
|
|
|
|
ClutterActor *stage = clutter_actor_get_stage (self);
|
|
|
|
|
|
|
|
/* do not queue a relayout on an unparented actor */
|
|
|
|
if (stage)
|
|
|
|
_clutter_stage_maybe_relayout (stage);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* commenting out the code above and just keeping this assigment
|
|
|
|
* implements 3)
|
|
|
|
*/
|
|
|
|
*box = self->priv->allocation;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_allocation_geometry:
|
|
|
|
* @self: A #ClutterActor
|
2009-02-26 19:42:19 +00:00
|
|
|
* @geom: (out): allocation geometry in pixels
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Gets the layout box an actor has been assigned. The allocation can
|
|
|
|
* only be assumed valid inside a paint() method; anywhere else, it
|
|
|
|
* may be out-of-date.
|
|
|
|
*
|
|
|
|
* An allocation does not incorporate the actor's scale or anchor point;
|
|
|
|
* those transformations do not affect layout, only rendering.
|
|
|
|
*
|
|
|
|
* The returned rectangle is in pixels.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_allocation_geometry (ClutterActor *self,
|
|
|
|
ClutterGeometry *geom)
|
|
|
|
{
|
2009-07-01 14:15:52 +00:00
|
|
|
ClutterActorBox box;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2009-07-01 14:15:52 +00:00
|
|
|
g_return_if_fail (geom != NULL);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2009-07-01 14:15:52 +00:00
|
|
|
clutter_actor_get_allocation_box (self, &box);
|
2009-04-15 16:25:50 +00:00
|
|
|
|
2009-07-01 14:15:52 +00:00
|
|
|
geom->x = clutter_actor_box_get_x (&box);
|
|
|
|
geom->y = clutter_actor_box_get_y (&box);
|
|
|
|
geom->width = clutter_actor_box_get_width (&box);
|
|
|
|
geom->height = clutter_actor_box_get_height (&box);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* clutter_actor_allocate:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* @box: new allocation of the actor, in parent-relative coordinates
|
2009-06-03 13:02:06 +00:00
|
|
|
* @flags: flags that control the allocation
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Called by the parent of an actor to assign the actor its size.
|
|
|
|
* Should never be called by applications (except when implementing
|
|
|
|
* a container or layout manager).
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Actors can know from their allocation box whether they have moved
|
2009-06-03 13:02:06 +00:00
|
|
|
* with respect to their parent actor. The @flags parameter describes
|
|
|
|
* additional information about the allocation, for instance whether
|
|
|
|
* the parent has moved with respect to the stage, for example because
|
|
|
|
* a grandparent's origin has moved.
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
2009-06-03 13:02:06 +00:00
|
|
|
clutter_actor_allocate (ClutterActor *self,
|
|
|
|
const ClutterActorBox *box,
|
|
|
|
ClutterAllocationFlags flags)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
ClutterActorClass *klass;
|
|
|
|
gboolean child_moved;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv = self->priv;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2008-06-11 17:12:29 +00:00
|
|
|
child_moved = (box->x1 != priv->allocation.x1 ||
|
|
|
|
box->y1 != priv->allocation.y1);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/* If we get an allocation "out of the blue"
|
|
|
|
* (we did not queue relayout), then we want to
|
|
|
|
* ignore it. But if we have needs_allocation set,
|
|
|
|
* we want to guarantee that allocate() virtual
|
|
|
|
* method is always called, i.e. that queue_relayout()
|
|
|
|
* always results in an allocate() invocation on
|
|
|
|
* an actor.
|
|
|
|
*
|
|
|
|
* The optimization here is to avoid re-allocating
|
|
|
|
* actors that did not queue relayout and were
|
|
|
|
* not moved.
|
|
|
|
*/
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (!priv->needs_allocation &&
|
2009-06-03 13:02:06 +00:00
|
|
|
!(flags & CLUTTER_ABSOLUTE_ORIGIN_CHANGED) &&
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
!child_moved &&
|
|
|
|
box->x2 == priv->allocation.x2 &&
|
|
|
|
box->y2 == priv->allocation.y2)
|
|
|
|
{
|
2009-04-15 16:25:50 +00:00
|
|
|
CLUTTER_NOTE (LAYOUT, "No allocation needed");
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
return;
|
|
|
|
}
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2009-06-03 13:02:06 +00:00
|
|
|
/* When ABSOLUTE_ORIGIN_CHANGED is passed in to
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* clutter_actor_allocate(), it indicates whether the parent has its
|
|
|
|
* absolute origin moved; when passed in to ClutterActor::allocate()
|
|
|
|
* virtual method though, it indicates whether the child has its
|
2009-06-03 13:02:06 +00:00
|
|
|
* absolute origin moved. So we set it when child_moved is TRUE
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*/
|
2009-06-03 13:02:06 +00:00
|
|
|
if (child_moved)
|
|
|
|
flags |= CLUTTER_ABSOLUTE_ORIGIN_CHANGED;
|
|
|
|
|
2009-10-23 09:17:40 +00:00
|
|
|
CLUTTER_SET_PRIVATE_FLAGS (self, CLUTTER_ACTOR_IN_RELAYOUT);
|
|
|
|
|
2009-06-03 13:02:06 +00:00
|
|
|
klass = CLUTTER_ACTOR_GET_CLASS (self);
|
|
|
|
klass->allocate (self, box, flags);
|
2009-10-23 09:17:40 +00:00
|
|
|
|
|
|
|
CLUTTER_UNSET_PRIVATE_FLAGS (self, CLUTTER_ACTOR_IN_RELAYOUT);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_set_geometry:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
* @geometry: A #ClutterGeometry
|
|
|
|
*
|
|
|
|
* Sets the actor's fixed position and forces its minimum and natural
|
|
|
|
* size, in pixels. This means the untransformed actor will have the
|
|
|
|
* given geometry. This is the same as calling clutter_actor_set_position()
|
|
|
|
* and clutter_actor_set_size().
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_geometry (ClutterActor *self,
|
|
|
|
const ClutterGeometry *geometry)
|
|
|
|
{
|
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
2009-04-15 16:25:50 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_set_position (self, geometry->x, geometry->y);
|
|
|
|
clutter_actor_set_size (self, geometry->width, geometry->height);
|
2009-04-15 16:25:50 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_geometry:
|
|
|
|
* @self: A #ClutterActor
|
2009-02-26 19:42:19 +00:00
|
|
|
* @geometry: (out): A location to store actors #ClutterGeometry
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Gets the size and position of an actor relative to its parent
|
|
|
|
* actor. This is the same as calling clutter_actor_get_position() and
|
|
|
|
* clutter_actor_get_size(). It tries to "do what you mean" and get the
|
|
|
|
* requested size and position if the actor's allocation is invalid.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_geometry (ClutterActor *self,
|
|
|
|
ClutterGeometry *geometry)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat x, y, width, height;
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_return_if_fail (geometry != NULL);
|
|
|
|
|
|
|
|
clutter_actor_get_position (self, &x, &y);
|
|
|
|
clutter_actor_get_size (self, &width, &height);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
geometry->x = (int) x;
|
|
|
|
geometry->y = (int) y;
|
|
|
|
geometry->width = (int) width;
|
|
|
|
geometry->height = (int) height;
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_set_position
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
* @x: New left position of actor in pixels.
|
|
|
|
* @y: New top position of actor in pixels.
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Sets the actor's fixed position in pixels relative to any parent
|
|
|
|
* actor.
|
|
|
|
*
|
|
|
|
* If a layout manager is in use, this position will override the
|
|
|
|
* layout manager and force a fixed position.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_position (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat x,
|
|
|
|
gfloat y)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2007-05-22 09:31:40 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
2009-04-15 16:25:50 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_set_x (self, x);
|
|
|
|
clutter_actor_set_y (self, y);
|
2009-04-15 16:25:50 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_fixed_position_set:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
|
|
|
* Checks whether an actor has a fixed position set (and will thus be
|
|
|
|
* unaffected by any layout manager).
|
|
|
|
*
|
|
|
|
* Return value: %TRUE if the fixed position is set on the actor
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_get_fixed_position_set (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), FALSE);
|
|
|
|
|
|
|
|
return self->priv->position_set;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_set_fixed_position_set:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
* @is_set: whether to use fixed position
|
|
|
|
*
|
|
|
|
* Sets whether an actor has a fixed position set (and will thus be
|
|
|
|
* unaffected by any layout manager).
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_fixed_position_set (ClutterActor *self,
|
|
|
|
gboolean is_set)
|
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
if (self->priv->position_set == (is_set != FALSE))
|
|
|
|
return;
|
|
|
|
|
|
|
|
self->priv->position_set = is_set != FALSE;
|
|
|
|
g_object_notify (G_OBJECT (self), "fixed-position-set");
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_queue_relayout (self);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_move_by:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
* @dx: Distance to move Actor on X axis.
|
|
|
|
* @dy: Distance to move Actor on Y axis.
|
|
|
|
*
|
|
|
|
* Moves an actor by the specified distance relative to its current
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* position in pixels.
|
|
|
|
*
|
|
|
|
* This function modifies the fixed position of an actor and thus removes
|
|
|
|
* it from any layout management. Another way to move an actor is with an
|
|
|
|
* anchor point, see clutter_actor_set_anchor_point().
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Since: 0.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_move_by (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat dx,
|
|
|
|
gfloat dy)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
gfloat x, y;
|
2009-04-15 16:25:50 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
x = priv->fixed_x;
|
|
|
|
y = priv->fixed_y;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_set_position (self, x + dx, y + dy);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_set_min_width (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat min_width)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2008-06-26 21:42:44 +00:00
|
|
|
/* if we are setting the size on a top-level actor and the
|
|
|
|
* backend only supports static top-levels (e.g. framebuffers)
|
|
|
|
* then we ignore the passed value and we override it with
|
|
|
|
* the stage implementation's preferred size.
|
2008-06-25 15:02:13 +00:00
|
|
|
*/
|
2008-06-26 21:42:44 +00:00
|
|
|
if ((CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL) &&
|
|
|
|
clutter_feature_available (CLUTTER_FEATURE_STAGE_STATIC))
|
|
|
|
return;
|
2008-06-25 14:47:00 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->min_width_set && min_width == priv->request_min_width)
|
|
|
|
return;
|
|
|
|
|
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
|
|
|
|
|
|
|
priv->request_min_width = min_width;
|
|
|
|
g_object_notify (G_OBJECT (self), "min-width");
|
|
|
|
clutter_actor_set_min_width_set (self, TRUE);
|
|
|
|
|
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_queue_relayout (self);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_set_min_height (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat min_height)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2008-06-26 21:42:44 +00:00
|
|
|
/* if we are setting the size on a top-level actor and the
|
|
|
|
* backend only supports static top-levels (e.g. framebuffers)
|
|
|
|
* then we ignore the passed value and we override it with
|
|
|
|
* the stage implementation's preferred size.
|
2008-06-25 15:02:13 +00:00
|
|
|
*/
|
2008-06-26 21:42:44 +00:00
|
|
|
if ((CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL) &&
|
|
|
|
clutter_feature_available (CLUTTER_FEATURE_STAGE_STATIC))
|
|
|
|
return;
|
2008-06-25 14:47:00 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->min_height_set && min_height == priv->request_min_height)
|
|
|
|
return;
|
|
|
|
|
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
|
|
|
|
|
|
|
priv->request_min_height = min_height;
|
|
|
|
g_object_notify (G_OBJECT (self), "min-height");
|
|
|
|
clutter_actor_set_min_height_set (self, TRUE);
|
|
|
|
|
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_queue_relayout (self);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_set_natural_width (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat natural_width)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2008-06-26 21:42:44 +00:00
|
|
|
/* if we are setting the size on a top-level actor and the
|
|
|
|
* backend only supports static top-levels (e.g. framebuffers)
|
|
|
|
* then we ignore the passed value and we override it with
|
|
|
|
* the stage implementation's preferred size.
|
2008-06-25 15:02:13 +00:00
|
|
|
*/
|
2008-06-26 21:42:44 +00:00
|
|
|
if ((CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL) &&
|
|
|
|
clutter_feature_available (CLUTTER_FEATURE_STAGE_STATIC))
|
|
|
|
return;
|
2008-06-25 14:47:00 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->natural_width_set &&
|
|
|
|
natural_width == priv->request_natural_width)
|
|
|
|
return;
|
|
|
|
|
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
|
|
|
|
|
|
|
priv->request_natural_width = natural_width;
|
|
|
|
g_object_notify (G_OBJECT (self), "natural-width");
|
|
|
|
clutter_actor_set_natural_width_set (self, TRUE);
|
|
|
|
|
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_queue_relayout (self);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_set_natural_height (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat natural_height)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2008-06-26 21:42:44 +00:00
|
|
|
/* if we are setting the size on a top-level actor and the
|
|
|
|
* backend only supports static top-levels (e.g. framebuffers)
|
|
|
|
* then we ignore the passed value and we override it with
|
|
|
|
* the stage implementation's preferred size.
|
2008-06-25 15:02:13 +00:00
|
|
|
*/
|
2008-06-26 21:42:44 +00:00
|
|
|
if ((CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL) &&
|
|
|
|
clutter_feature_available (CLUTTER_FEATURE_STAGE_STATIC))
|
|
|
|
return;
|
2008-06-25 14:47:00 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->natural_height_set &&
|
|
|
|
natural_height == priv->request_natural_height)
|
|
|
|
return;
|
|
|
|
|
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
|
|
|
|
|
|
|
priv->request_natural_height = natural_height;
|
|
|
|
g_object_notify (G_OBJECT (self), "natural-height");
|
|
|
|
clutter_actor_set_natural_height_set (self, TRUE);
|
|
|
|
|
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_queue_relayout (self);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_set_min_width_set (ClutterActor *self,
|
|
|
|
gboolean use_min_width)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->min_width_set == (use_min_width != FALSE))
|
|
|
|
return;
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->min_width_set = use_min_width != FALSE;
|
|
|
|
g_object_notify (G_OBJECT (self), "min-width-set");
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_queue_relayout (self);
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_set_min_height_set (ClutterActor *self,
|
|
|
|
gboolean use_min_height)
|
2008-01-04 18:30:25 +00:00
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
2008-01-04 18:30:25 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->min_height_set == (use_min_height != FALSE))
|
|
|
|
return;
|
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
|
|
|
|
|
|
|
priv->min_height_set = use_min_height != FALSE;
|
|
|
|
g_object_notify (G_OBJECT (self), "min-height-set");
|
|
|
|
|
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
|
|
|
|
|
|
|
clutter_actor_queue_relayout (self);
|
2008-01-04 18:30:25 +00:00
|
|
|
}
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_set_natural_width_set (ClutterActor *self,
|
|
|
|
gboolean use_natural_width)
|
2006-11-15 21:19:01 +00:00
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->natural_width_set == (use_natural_width != FALSE))
|
|
|
|
return;
|
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
2006-11-15 21:19:01 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->natural_width_set = use_natural_width != FALSE;
|
|
|
|
g_object_notify (G_OBJECT (self), "natural-width-set");
|
2006-11-15 21:19:01 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
2006-11-15 21:19:01 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_queue_relayout (self);
|
2006-11-15 21:19:01 +00:00
|
|
|
}
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_set_natural_height_set (ClutterActor *self,
|
|
|
|
gboolean use_natural_height)
|
2007-11-15 17:35:47 +00:00
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
ClutterActorPrivate *priv = self->priv;
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
2007-11-15 17:35:47 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->natural_height_set == (use_natural_height != FALSE))
|
|
|
|
return;
|
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
2007-11-15 17:35:47 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->natural_height_set = use_natural_height != FALSE;
|
|
|
|
g_object_notify (G_OBJECT (self), "natural-height-set");
|
2007-11-15 17:35:47 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
2007-11-15 17:35:47 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_queue_relayout (self);
|
2007-11-15 17:35:47 +00:00
|
|
|
}
|
|
|
|
|
2009-10-07 14:15:02 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_request_mode:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @mode: the request mode
|
|
|
|
*
|
|
|
|
* Sets the geometry request mode of @self.
|
|
|
|
*
|
|
|
|
* The @mode determines the order for invoking
|
|
|
|
* clutter_actor_get_preferred_width() and
|
|
|
|
* clutter_actor_get_preferred_height()
|
|
|
|
*
|
|
|
|
* Since: 1.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_request_mode (ClutterActor *self,
|
|
|
|
ClutterRequestMode mode)
|
2007-11-28 15:31:41 +00:00
|
|
|
{
|
2009-10-07 14:15:02 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
if (priv->request_mode == mode)
|
|
|
|
return;
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->request_mode = mode;
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->needs_width_request = TRUE;
|
|
|
|
priv->needs_height_request = TRUE;
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_object_notify (G_OBJECT (self), "request-mode");
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_queue_relayout (self);
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2009-10-07 14:15:02 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_request_mode:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the geometry request mode of @self
|
|
|
|
*
|
|
|
|
* Return value: the request mode for the actor
|
|
|
|
*
|
|
|
|
* Since: 1.2
|
|
|
|
*/
|
|
|
|
ClutterRequestMode
|
|
|
|
clutter_actor_get_request_mode (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self),
|
|
|
|
CLUTTER_REQUEST_HEIGHT_FOR_WIDTH);
|
|
|
|
|
|
|
|
return self->priv->request_mode;
|
|
|
|
}
|
|
|
|
|
2009-08-13 11:26:09 +00:00
|
|
|
/* variant of set_width() without checks and without notification
|
|
|
|
* freeze+thaw, for internal usage only
|
|
|
|
*/
|
|
|
|
static inline void
|
|
|
|
clutter_actor_set_width_internal (ClutterActor *self,
|
|
|
|
gfloat width)
|
|
|
|
{
|
|
|
|
if (width >= 0)
|
|
|
|
{
|
|
|
|
clutter_actor_set_min_width (self, width);
|
|
|
|
clutter_actor_set_natural_width (self, width);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
clutter_actor_set_min_width_set (self, FALSE);
|
|
|
|
clutter_actor_set_natural_width_set (self, FALSE);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* variant of set_height() without checks and without notification
|
|
|
|
* freeze+thaw, for internal usage only
|
|
|
|
*/
|
|
|
|
static inline void
|
|
|
|
clutter_actor_set_height_internal (ClutterActor *self,
|
|
|
|
gfloat height)
|
|
|
|
{
|
|
|
|
if (height >= 0)
|
|
|
|
{
|
|
|
|
clutter_actor_set_min_height (self, height);
|
|
|
|
clutter_actor_set_natural_height (self, height);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
clutter_actor_set_min_height_set (self, FALSE);
|
|
|
|
clutter_actor_set_natural_height_set (self, FALSE);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_size
|
|
|
|
* @self: A #ClutterActor
|
2007-11-15 17:35:47 +00:00
|
|
|
* @width: New width of actor in pixels, or -1
|
|
|
|
* @height: New height of actor in pixels, or -1
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Sets the actor's size request in pixels. This overrides any
|
|
|
|
* "normal" size request the actor would have. For example
|
|
|
|
* a text actor might normally request the size of the text;
|
|
|
|
* this function would force a specific size instead.
|
2008-02-07 17:08:29 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* If @width and/or @height are -1 the actor will use its
|
|
|
|
* "normal" size request instead of overriding it, i.e.
|
|
|
|
* you can "unset" the size with -1.
|
2008-02-07 17:08:29 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* This function sets or unsets both the minimum and natural size.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_size (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat width,
|
|
|
|
gfloat height)
|
2007-11-28 15:31:41 +00:00
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
2009-08-13 11:26:09 +00:00
|
|
|
clutter_actor_set_width_internal (self, width);
|
|
|
|
clutter_actor_set_height_internal (self, height);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2006-12-13 18:12:09 +00:00
|
|
|
/**
|
2006-12-12 20:20:04 +00:00
|
|
|
* clutter_actor_get_size:
|
2006-11-15 21:19:01 +00:00
|
|
|
* @self: A #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @width: (out) (allow-none): return location for the width, or %NULL.
|
|
|
|
* @height: (out) (allow-none): return location for the height, or %NULL.
|
2006-11-15 21:19:01 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* This function tries to "do what you mean" and return
|
|
|
|
* the size an actor will have. If the actor has a valid
|
|
|
|
* allocation, the allocation will be returned; otherwise,
|
|
|
|
* the actors natural size request will be returned.
|
|
|
|
*
|
|
|
|
* If you care whether you get the request vs. the allocation, you
|
|
|
|
* should probably call a different function like
|
2009-07-01 14:15:52 +00:00
|
|
|
* clutter_actor_get_allocation_box() or
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* clutter_actor_get_preferred_width().
|
2006-11-15 21:19:01 +00:00
|
|
|
*
|
|
|
|
* Since: 0.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_size (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat *width,
|
|
|
|
gfloat *height)
|
2006-11-15 21:19:01 +00:00
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
if (width)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*width = clutter_actor_get_width (self);
|
2006-11-20 12:02:24 +00:00
|
|
|
|
2006-11-15 21:19:01 +00:00
|
|
|
if (height)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*height = clutter_actor_get_height (self);
|
2006-11-15 21:19:01 +00:00
|
|
|
}
|
|
|
|
|
2007-11-15 17:35:47 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_position:
|
|
|
|
* @self: a #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @x: (out) (allow-none): return location for the X coordinate, or %NULL
|
|
|
|
* @y: (out) (allow-none): return location for the Y coordinate, or %NULL
|
2007-11-15 17:35:47 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* This function tries to "do what you mean" and tell you where the
|
|
|
|
* actor is, prior to any transformations. Retrieves the fixed
|
|
|
|
* position of an actor in pixels, if one has been set; otherwise, if
|
|
|
|
* the allocation is valid, returns the actor's allocated position;
|
|
|
|
* otherwise, returns 0,0.
|
|
|
|
*
|
|
|
|
* The returned position is in pixels.
|
2007-11-15 17:35:47 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_position (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat *x,
|
|
|
|
gfloat *y)
|
2007-11-15 17:35:47 +00:00
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
if (x)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*x = clutter_actor_get_x (self);
|
2007-11-15 17:35:47 +00:00
|
|
|
|
|
|
|
if (y)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*y = clutter_actor_get_y (self);
|
2007-11-15 17:35:47 +00:00
|
|
|
}
|
|
|
|
|
2007-12-07 15:49:54 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_get_transformed_position:
|
|
|
|
* @self: A #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @x: (out) (allow-none): return location for the X coordinate, or %NULL
|
|
|
|
* @y: (out) (allow-none): return location for the Y coordinate, or %NULL
|
2007-12-07 15:49:54 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Gets the absolute position of an actor, in pixels relative to the stage.
|
2007-05-22 09:31:40 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Since: 0.8
|
2007-05-22 09:31:40 +00:00
|
|
|
*/
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_get_transformed_position (ClutterActor *self,
|
|
|
|
gfloat *x,
|
|
|
|
gfloat *y)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterVertex v1;
|
|
|
|
ClutterVertex v2;
|
2007-07-28 17:11:39 +00:00
|
|
|
|
2009-06-05 21:53:57 +00:00
|
|
|
v1.x = v1.y = v1.z = 0;
|
2007-07-02 09:21:58 +00:00
|
|
|
clutter_actor_apply_transform_to_point (self, &v1, &v2);
|
|
|
|
|
2007-07-28 17:11:39 +00:00
|
|
|
if (x)
|
|
|
|
*x = v2.x;
|
2007-05-23 10:47:35 +00:00
|
|
|
|
2007-07-28 17:11:39 +00:00
|
|
|
if (y)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*y = v2.y;
|
2007-05-18 07:30:06 +00:00
|
|
|
}
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_get_transformed_size:
|
2006-08-29 19:09:43 +00:00
|
|
|
* @self: A #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @width: (out) (allow-none): return location for the width, or %NULL
|
|
|
|
* @height: (out) (allow-none): return location for the height, or %NULL
|
2008-02-15 14:39:25 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Gets the absolute size of an actor in pixels, taking into account the
|
|
|
|
* scaling factors.
|
2008-02-15 14:39:25 +00:00
|
|
|
*
|
2008-07-01 21:52:19 +00:00
|
|
|
* If the actor has a valid allocation, the allocated size will be used.
|
|
|
|
* If the actor has not a valid allocation then the preferred size will
|
|
|
|
* be transformed and returned.
|
|
|
|
*
|
|
|
|
* If you want the transformed allocation, see
|
|
|
|
* clutter_actor_get_abs_allocation_vertices() instead.
|
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* <note>When the actor (or one of its ancestors) is rotated around the
|
|
|
|
* X or Y axis, it no longer appears as on the stage as a rectangle, but
|
|
|
|
* as a generic quadrangle; in that case this function returns the size
|
|
|
|
* of the smallest rectangle that encapsulates the entire quad. Please
|
|
|
|
* note that in this case no assumptions can be made about the relative
|
|
|
|
* position of this envelope to the absolute position of the actor, as
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* returned by clutter_actor_get_transformed_position(); if you need this
|
|
|
|
* information, you need to use clutter_actor_get_abs_allocation_vertices()
|
|
|
|
* to get the coords of the actual quadrangle.</note>
|
2007-06-14 08:17:13 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Since: 0.8
|
2006-08-29 19:09:43 +00:00
|
|
|
*/
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_get_transformed_size (ClutterActor *self,
|
|
|
|
gfloat *width,
|
|
|
|
gfloat *height)
|
2006-08-29 19:09:43 +00:00
|
|
|
{
|
2008-07-01 21:52:19 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2007-06-14 08:17:13 +00:00
|
|
|
ClutterVertex v[4];
|
2009-02-23 12:55:01 +00:00
|
|
|
gfloat x_min, x_max, y_min, y_max;
|
2007-06-14 08:17:13 +00:00
|
|
|
gint i;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-07-01 21:52:19 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
/* if the actor hasn't been allocated yet, get the preferred
|
|
|
|
* size and transform that
|
|
|
|
*/
|
|
|
|
if (priv->needs_allocation)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat natural_width, natural_height;
|
2008-07-01 21:52:19 +00:00
|
|
|
ClutterActorBox box;
|
|
|
|
|
|
|
|
/* make a fake allocation to transform */
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_get_position (self, &box.x1, &box.y1);
|
2008-07-01 21:52:19 +00:00
|
|
|
|
|
|
|
natural_width = natural_height = 0;
|
|
|
|
clutter_actor_get_preferred_size (self, NULL, NULL,
|
|
|
|
&natural_width,
|
|
|
|
&natural_height);
|
|
|
|
|
|
|
|
box.x2 = box.x1 + natural_width;
|
|
|
|
box.y2 = box.y1 + natural_height;
|
2009-02-18 18:54:54 +00:00
|
|
|
|
2008-07-01 21:52:19 +00:00
|
|
|
clutter_actor_transform_and_project_box (self, &box, v);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
clutter_actor_get_abs_allocation_vertices (self, v);
|
2006-08-29 19:09:43 +00:00
|
|
|
|
2007-06-14 08:17:13 +00:00
|
|
|
x_min = x_max = v[0].x;
|
|
|
|
y_min = y_max = v[0].y;
|
2006-08-29 19:09:43 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
for (i = 1; i < G_N_ELEMENTS (v); ++i)
|
2006-08-29 19:09:43 +00:00
|
|
|
{
|
2007-06-14 08:17:13 +00:00
|
|
|
if (v[i].x < x_min)
|
|
|
|
x_min = v[i].x;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-06-14 08:17:13 +00:00
|
|
|
if (v[i].x > x_max)
|
|
|
|
x_max = v[i].x;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-06-14 08:17:13 +00:00
|
|
|
if (v[i].y < y_min)
|
|
|
|
y_min = v[i].y;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-06-14 08:17:13 +00:00
|
|
|
if (v[i].y > y_max)
|
|
|
|
y_max = v[i].y;
|
2007-05-18 07:30:06 +00:00
|
|
|
}
|
2007-06-14 08:17:13 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (width)
|
|
|
|
*width = x_max - x_min;
|
|
|
|
|
|
|
|
if (height)
|
|
|
|
*height = y_max - y_min;
|
2007-05-18 07:30:06 +00:00
|
|
|
}
|
|
|
|
|
2007-05-23 10:47:35 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_get_width:
|
2007-05-23 10:47:35 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Retrieves the width of a #ClutterActor.
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* If the actor has a valid allocation, this function will return the
|
2008-06-12 16:20:08 +00:00
|
|
|
* width of the allocated area given to the actor.
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* If the actor does not have a valid allocation, this function will
|
2008-06-12 16:20:08 +00:00
|
|
|
* return the actor's natural width, that is the preferred width of
|
|
|
|
* the actor.
|
|
|
|
*
|
|
|
|
* If you care whether you get the preferred width or the width that
|
|
|
|
* has been assigned to the actor, you should probably call a different
|
2009-07-01 14:15:52 +00:00
|
|
|
* function like clutter_actor_get_allocation_box() to retrieve the
|
2008-06-12 16:20:08 +00:00
|
|
|
* allocated size or clutter_actor_get_preferred_width() to retrieve the
|
|
|
|
* preferred width.
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
2008-06-12 16:20:08 +00:00
|
|
|
* If an actor has a fixed width, for instance a width that has been
|
|
|
|
* assigned using clutter_actor_set_width(), the width returned will
|
|
|
|
* be the same value.
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* Return value: the width of the actor, in pixels
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_get_width (ClutterActor *self)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-11-28 15:31:41 +00:00
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0);
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->needs_allocation)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat natural_width = 0;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
if (self->priv->request_mode == CLUTTER_REQUEST_HEIGHT_FOR_WIDTH)
|
|
|
|
clutter_actor_get_preferred_width (self, -1, NULL, &natural_width);
|
|
|
|
else
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat natural_height = 0;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
clutter_actor_get_preferred_height (self, -1, NULL, &natural_height);
|
|
|
|
clutter_actor_get_preferred_width (self, natural_height,
|
|
|
|
NULL,
|
|
|
|
&natural_width);
|
|
|
|
}
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
return natural_width;
|
|
|
|
}
|
|
|
|
else
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
return priv->allocation.x2 - priv->allocation.x1;
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_get_height:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Retrieves the height of a #ClutterActor.
|
2007-10-12 08:17:00 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* If the actor has a valid allocation, this function will return the
|
2008-06-12 16:20:08 +00:00
|
|
|
* height of the allocated area given to the actor.
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
|
|
|
* If the actor does not have a valid allocation, this function will
|
2008-06-12 16:20:08 +00:00
|
|
|
* return the actor's natural height, that is the preferred height of
|
|
|
|
* the actor.
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
2008-06-12 16:20:08 +00:00
|
|
|
* If you care whether you get the preferred height or the height that
|
|
|
|
* has been assigned to the actor, you should probably call a different
|
2009-07-01 14:15:52 +00:00
|
|
|
* function like clutter_actor_get_allocation_box() to retrieve the
|
2008-06-12 16:20:08 +00:00
|
|
|
* allocated size or clutter_actor_get_preferred_height() to retrieve the
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* preferred height.
|
|
|
|
*
|
2008-06-12 16:20:08 +00:00
|
|
|
* If an actor has a fixed height, for instance a height that has been
|
|
|
|
* assigned using clutter_actor_set_height(), the height returned will
|
|
|
|
* be the same value.
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Return value: the height of the actor, in pixels
|
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_get_height (ClutterActor *self)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-11-28 15:31:41 +00:00
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0);
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->needs_allocation)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat natural_height = 0;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (priv->request_mode == CLUTTER_REQUEST_HEIGHT_FOR_WIDTH)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat natural_width = 0;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
clutter_actor_get_preferred_width (self, -1, NULL, &natural_width);
|
|
|
|
clutter_actor_get_preferred_height (self, natural_width,
|
|
|
|
NULL, &natural_height);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
clutter_actor_get_preferred_height (self, -1, NULL, &natural_height);
|
2007-11-28 15:31:41 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
return natural_height;
|
|
|
|
}
|
|
|
|
else
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
return priv->allocation.y2 - priv->allocation.y1;
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2007-01-04 19:56:01 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_width
|
|
|
|
* @self: A #ClutterActor
|
2009-08-13 11:26:09 +00:00
|
|
|
* @width: Requested new width for the actor, in pixels, or -1
|
2007-10-12 08:17:00 +00:00
|
|
|
*
|
2008-06-12 16:20:08 +00:00
|
|
|
* Forces a width on an actor, causing the actor's preferred width
|
|
|
|
* and height (if any) to be ignored.
|
|
|
|
*
|
2009-08-13 11:26:09 +00:00
|
|
|
* If @width is -1 the actor will use its preferred width request
|
|
|
|
* instead of overriding it, i.e. you can "unset" the width with -1.
|
|
|
|
*
|
2008-06-12 16:20:08 +00:00
|
|
|
* This function sets both the minimum and natural size of the actor.
|
2007-01-04 19:56:01 +00:00
|
|
|
*
|
2007-11-28 15:31:41 +00:00
|
|
|
* since: 0.2
|
2009-08-13 11:26:09 +00:00
|
|
|
*/
|
2007-01-04 19:56:01 +00:00
|
|
|
void
|
2007-11-15 17:35:47 +00:00
|
|
|
clutter_actor_set_width (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat width)
|
2007-11-28 15:31:41 +00:00
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
2009-08-13 11:26:09 +00:00
|
|
|
clutter_actor_set_width_internal (self, width);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2007-01-04 19:56:01 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_height
|
|
|
|
* @self: A #ClutterActor
|
2009-08-13 11:26:09 +00:00
|
|
|
* @height: Requested new height for the actor, in pixels, or -1
|
2007-10-12 08:17:00 +00:00
|
|
|
*
|
2008-06-12 16:20:08 +00:00
|
|
|
* Forces a height on an actor, causing the actor's preferred width
|
|
|
|
* and height (if any) to be ignored.
|
|
|
|
*
|
2009-08-13 11:26:09 +00:00
|
|
|
* If @height is -1 the actor will use its preferred height instead of
|
|
|
|
* overriding it, i.e. you can "unset" the height with -1.
|
|
|
|
*
|
2008-06-12 16:20:08 +00:00
|
|
|
* This function sets both the minimum and natural size of the actor.
|
2007-01-04 19:56:01 +00:00
|
|
|
*
|
2007-11-28 15:31:41 +00:00
|
|
|
* since: 0.2
|
2009-08-13 11:26:09 +00:00
|
|
|
*/
|
2007-01-04 19:56:01 +00:00
|
|
|
void
|
2007-11-15 17:35:47 +00:00
|
|
|
clutter_actor_set_height (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat height)
|
2007-11-28 15:31:41 +00:00
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
2009-08-13 11:26:09 +00:00
|
|
|
clutter_actor_set_height_internal (self, height);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2007-11-15 17:35:47 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_x:
|
|
|
|
* @self: a #ClutterActor
|
2008-02-15 14:39:25 +00:00
|
|
|
* @x: the actor's position on the X axis
|
2007-11-15 17:35:47 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Sets the actor's X coordinate, relative to its parent, in pixels.
|
2008-06-12 16:20:08 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Overrides any layout manager and forces a fixed position for
|
|
|
|
* the actor.
|
2007-11-15 17:35:47 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_x (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat x)
|
2007-11-28 15:31:41 +00:00
|
|
|
{
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2007-11-28 15:31:41 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->position_set && priv->fixed_x == x)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
return;
|
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv->fixed_x = x;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_set_fixed_position_set (self, TRUE);
|
|
|
|
|
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
|
|
|
|
|
|
|
clutter_actor_queue_relayout (self);
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2007-11-15 17:35:47 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_y:
|
|
|
|
* @self: a #ClutterActor
|
2008-02-15 14:39:25 +00:00
|
|
|
* @y: the actor's position on the Y axis
|
2007-11-15 17:35:47 +00:00
|
|
|
*
|
2008-06-12 16:20:08 +00:00
|
|
|
* Sets the actor's Y coordinate, relative to its parent, in pixels.#
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Overrides any layout manager and forces a fixed position for
|
|
|
|
* the actor.
|
2007-11-15 17:35:47 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_y (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat y)
|
2007-11-28 15:31:41 +00:00
|
|
|
{
|
2008-06-11 15:08:03 +00:00
|
|
|
ClutterActorBox old = { 0, };
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2007-11-28 15:31:41 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->position_set && priv->fixed_y == y)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
return;
|
|
|
|
|
|
|
|
clutter_actor_store_old_geometry (self, &old);
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv->fixed_y = y;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_set_fixed_position_set (self, TRUE);
|
|
|
|
|
|
|
|
clutter_actor_notify_if_geometry_changed (self, &old);
|
|
|
|
|
|
|
|
clutter_actor_queue_relayout (self);
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_x
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Retrieves the X coordinate of a #ClutterActor.
|
|
|
|
*
|
|
|
|
* This function tries to "do what you mean", by returning the
|
|
|
|
* correct value depending on the actor's state.
|
|
|
|
*
|
|
|
|
* If the actor has a valid allocation, this function will return
|
|
|
|
* the X coordinate of the origin of the allocation box.
|
|
|
|
*
|
|
|
|
* If the actor has any fixed coordinate set using clutter_actor_set_x(),
|
|
|
|
* clutter_actor_set_position() or clutter_actor_set_geometry(), this
|
|
|
|
* function will return that coordinate.
|
|
|
|
*
|
|
|
|
* If both the allocation and a fixed position are missing, this function
|
|
|
|
* will return 0.
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* Return value: the X coordinate, in pixels, ignoring any
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* transformation (i.e. scaling, rotation)
|
2008-02-15 14:39:25 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_get_x (ClutterActor *self)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2007-05-18 07:30:06 +00:00
|
|
|
|
2007-11-28 15:31:41 +00:00
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0);
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->needs_allocation)
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (priv->position_set)
|
|
|
|
return priv->fixed_x;
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
else
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
else
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
return priv->allocation.x1;
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* clutter_actor_get_y
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Retrieves the Y coordinate of a #ClutterActor.
|
|
|
|
*
|
|
|
|
* This function tries to "do what you mean", by returning the
|
|
|
|
* correct value depending on the actor's state.
|
|
|
|
*
|
|
|
|
* If the actor has a valid allocation, this function will return
|
|
|
|
* the Y coordinate of the origin of the allocation box.
|
|
|
|
*
|
|
|
|
* If the actor has any fixed coordinate set using clutter_actor_set_y(),
|
|
|
|
* clutter_actor_set_position() or clutter_actor_set_geometry(), this
|
|
|
|
* function will return that coordinate.
|
|
|
|
*
|
|
|
|
* If both the allocation and a fixed position are missing, this function
|
|
|
|
* will return 0.
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* Return value: the Y coordinate, in pixels, ignoring any
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* transformation (i.e. scaling, rotation)
|
2008-02-15 14:39:25 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_get_y (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (self->priv->needs_allocation)
|
|
|
|
{
|
|
|
|
if (self->priv->position_set)
|
|
|
|
return self->priv->fixed_y;
|
|
|
|
else
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
return self->priv->allocation.y1;
|
2007-11-28 15:31:41 +00:00
|
|
|
}
|
|
|
|
|
2006-08-29 19:09:43 +00:00
|
|
|
/**
|
2009-01-23 11:11:24 +00:00
|
|
|
* clutter_actor_set_scale:
|
2006-08-29 19:09:43 +00:00
|
|
|
* @self: A #ClutterActor
|
2009-01-23 11:11:24 +00:00
|
|
|
* @scale_x: double factor to scale actor by horizontally.
|
|
|
|
* @scale_y: double factor to scale actor by vertically.
|
2006-08-29 19:09:43 +00:00
|
|
|
*
|
2009-01-23 17:25:43 +00:00
|
|
|
* Scales an actor with the given factors. The scaling is relative to
|
|
|
|
* the scale center and the anchor point. The scale center is
|
|
|
|
* unchanged by this function and defaults to 0,0.
|
2008-02-15 14:39:25 +00:00
|
|
|
*
|
2009-01-23 11:11:24 +00:00
|
|
|
* Since: 0.2
|
2006-08-29 19:09:43 +00:00
|
|
|
*/
|
|
|
|
void
|
2009-01-23 11:11:24 +00:00
|
|
|
clutter_actor_set_scale (ClutterActor *self,
|
|
|
|
gdouble scale_x,
|
|
|
|
gdouble scale_y)
|
2006-08-29 19:09:43 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
2006-08-29 19:09:43 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
2007-09-27 21:38:38 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv->scale_x = scale_x;
|
2007-09-27 21:38:38 +00:00
|
|
|
g_object_notify (G_OBJECT (self), "scale-x");
|
2007-10-12 08:17:00 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv->scale_y = scale_y;
|
2007-09-27 21:38:38 +00:00
|
|
|
g_object_notify (G_OBJECT (self), "scale-y");
|
|
|
|
|
2006-08-29 19:09:43 +00:00
|
|
|
if (CLUTTER_ACTOR_IS_VISIBLE (self))
|
|
|
|
clutter_actor_queue_redraw (self);
|
2009-01-23 11:11:24 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2006-08-29 19:09:43 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2009-01-23 17:25:43 +00:00
|
|
|
* clutter_actor_set_scale_full:
|
2006-08-29 19:09:43 +00:00
|
|
|
* @self: A #ClutterActor
|
2007-06-22 15:16:40 +00:00
|
|
|
* @scale_x: double factor to scale actor by horizontally.
|
|
|
|
* @scale_y: double factor to scale actor by vertically.
|
2009-01-23 17:25:43 +00:00
|
|
|
* @center_x: X coordinate of the center of the scale.
|
|
|
|
* @center_y: Y coordinate of the center of the scale
|
2006-12-13 18:12:09 +00:00
|
|
|
*
|
2009-01-23 17:25:43 +00:00
|
|
|
* Scales an actor with the given factors around the given center
|
|
|
|
* point. The center point is specified in pixels relative to the
|
|
|
|
* anchor point (usually the top left corner of the actor).
|
2006-12-13 18:12:09 +00:00
|
|
|
*
|
2009-01-23 17:25:43 +00:00
|
|
|
* Since: 1.0
|
2006-08-29 19:09:43 +00:00
|
|
|
*/
|
|
|
|
void
|
2009-01-23 11:11:24 +00:00
|
|
|
clutter_actor_set_scale_full (ClutterActor *self,
|
|
|
|
gdouble scale_x,
|
|
|
|
gdouble scale_y,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat center_x,
|
|
|
|
gfloat center_y)
|
2006-08-29 19:09:43 +00:00
|
|
|
{
|
2009-01-23 11:11:24 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
priv = self->priv;
|
2006-08-29 19:09:43 +00:00
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_set_scale (self, scale_x, scale_y);
|
|
|
|
|
|
|
|
if (priv->scale_center.is_fractional)
|
|
|
|
g_object_notify (G_OBJECT (self), "scale-gravity");
|
2009-02-23 12:55:01 +00:00
|
|
|
|
2009-01-23 11:11:24 +00:00
|
|
|
g_object_notify (G_OBJECT (self), "scale-center-x");
|
|
|
|
g_object_notify (G_OBJECT (self), "scale-center-y");
|
|
|
|
|
|
|
|
clutter_anchor_coord_set_units (&priv->scale_center, center_x, center_y, 0);
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
}
|
|
|
|
|
2009-01-23 17:25:43 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_scale_with_gravity:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
* @scale_x: double factor to scale actor by horizontally.
|
|
|
|
* @scale_y: double factor to scale actor by vertically.
|
|
|
|
* @gravity: the location of the scale center expressed as a compass
|
|
|
|
* direction.
|
|
|
|
*
|
|
|
|
* Scales an actor with the given factors around the given
|
|
|
|
* center point. The center point is specified as one of the compass
|
|
|
|
* directions in #ClutterGravity. For example, setting it to north
|
|
|
|
* will cause the top of the actor to remain unchanged and the rest of
|
|
|
|
* the actor to expand left, right and downwards.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
2009-01-23 11:11:24 +00:00
|
|
|
void
|
|
|
|
clutter_actor_set_scale_with_gravity (ClutterActor *self,
|
|
|
|
gdouble scale_x,
|
|
|
|
gdouble scale_y,
|
|
|
|
ClutterGravity gravity)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (gravity == CLUTTER_GRAVITY_NONE)
|
|
|
|
clutter_actor_set_scale_full (self, scale_x, scale_y, 0, 0);
|
|
|
|
else
|
|
|
|
{
|
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_set_scale (self, scale_x, scale_y);
|
|
|
|
|
|
|
|
g_object_notify (G_OBJECT (self), "scale-gravity");
|
|
|
|
g_object_notify (G_OBJECT (self), "scale-center-x");
|
|
|
|
g_object_notify (G_OBJECT (self), "scale-center-y");
|
|
|
|
|
|
|
|
clutter_anchor_coord_set_gravity (&priv->scale_center, gravity);
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
}
|
2006-08-29 19:09:43 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2006-12-12 20:20:04 +00:00
|
|
|
* clutter_actor_get_scale:
|
2006-08-29 19:09:43 +00:00
|
|
|
* @self: A #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @scale_x: (out) (allow-none): Location to store horizonal
|
|
|
|
* scale factor, or %NULL.
|
|
|
|
* @scale_y: (out) (allow-none): Location to store vertical
|
|
|
|
* scale factor, or %NULL.
|
2006-08-29 19:09:43 +00:00
|
|
|
*
|
2009-07-02 13:21:16 +00:00
|
|
|
* Retrieves an actors scale factors.
|
2006-12-12 20:20:04 +00:00
|
|
|
*
|
|
|
|
* Since: 0.2
|
2006-08-29 19:09:43 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_scale (ClutterActor *self,
|
2006-11-20 12:02:24 +00:00
|
|
|
gdouble *scale_x,
|
|
|
|
gdouble *scale_y)
|
2006-08-29 19:09:43 +00:00
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2006-08-29 19:09:43 +00:00
|
|
|
if (scale_x)
|
2009-02-23 12:55:01 +00:00
|
|
|
*scale_x = self->priv->scale_x;
|
2006-08-29 19:09:43 +00:00
|
|
|
|
|
|
|
if (scale_y)
|
2009-02-23 12:55:01 +00:00
|
|
|
*scale_y = self->priv->scale_y;
|
2006-08-29 19:09:43 +00:00
|
|
|
}
|
|
|
|
|
2009-01-23 17:25:43 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_scale_center:
|
|
|
|
* @self: A #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @center_x: (out) (allow-none): Location to store the X position
|
|
|
|
* of the scale center, or %NULL.
|
|
|
|
* @center_y: (out) (allow-none): Location to store the Y position
|
|
|
|
* of the scale center, or %NULL.
|
2009-01-23 17:25:43 +00:00
|
|
|
*
|
|
|
|
* Retrieves the scale center coordinate in pixels relative to the top
|
|
|
|
* left corner of the actor. If the scale center was specified using a
|
|
|
|
* #ClutterGravity this will calculate the pixel offset using the
|
|
|
|
* current size of the actor.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
2009-01-23 11:11:24 +00:00
|
|
|
void
|
|
|
|
clutter_actor_get_scale_center (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat *center_x,
|
|
|
|
gfloat *center_y)
|
2009-01-23 11:11:24 +00:00
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
clutter_anchor_coord_get_units (self, &self->priv->scale_center,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
center_x,
|
|
|
|
center_y,
|
|
|
|
NULL);
|
2009-01-23 11:11:24 +00:00
|
|
|
}
|
|
|
|
|
2009-01-23 17:25:43 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_scale_gravity:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the scale center as a compass direction. If the scale
|
|
|
|
* center was specified in pixels or units this will return
|
|
|
|
* %CLUTTER_GRAVITY_NONE.
|
|
|
|
*
|
|
|
|
* Return value: the scale gravity
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
2009-01-23 11:11:24 +00:00
|
|
|
ClutterGravity
|
|
|
|
clutter_actor_get_scale_gravity (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), CLUTTER_GRAVITY_NONE);
|
|
|
|
|
|
|
|
return clutter_anchor_coord_get_gravity (&self->priv->scale_center);
|
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_opacity:
|
|
|
|
* @self: A #ClutterActor
|
2008-02-15 15:18:02 +00:00
|
|
|
* @opacity: New opacity value for the actor.
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Sets the actor's opacity, with zero being completely transparent and
|
|
|
|
* 255 (0xff) being fully opaque.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_opacity (ClutterActor *self,
|
2007-05-22 09:31:40 +00:00
|
|
|
guint8 opacity)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2007-10-12 08:17:00 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
priv = self->priv;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (priv->opacity != opacity)
|
|
|
|
{
|
|
|
|
priv->opacity = opacity;
|
|
|
|
|
|
|
|
clutter_actor_queue_redraw (self);
|
|
|
|
|
|
|
|
g_object_notify (G_OBJECT (self), "opacity");
|
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* clutter_actor_get_paint_opacity:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2008-02-20 10:59:47 +00:00
|
|
|
* Retrieves the absolute opacity of the actor, as it appears on the stage.
|
|
|
|
*
|
|
|
|
* This function traverses the hierarchy chain and composites the opacity of
|
|
|
|
* the actor with that of its parents.
|
|
|
|
*
|
|
|
|
* This function is intended for subclasses to use in the paint virtual
|
|
|
|
* function, to paint themselves with the correct opacity.
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
|
|
|
* Return value: The actor opacity value.
|
2008-02-20 10:59:47 +00:00
|
|
|
*
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* Since: 0.8
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
guint8
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_get_paint_opacity (ClutterActor *self)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2008-02-20 10:59:47 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2006-06-13 13:17:45 +00:00
|
|
|
ClutterActor *parent;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0);
|
|
|
|
|
2008-02-20 10:59:47 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
2009-01-17 16:51:03 +00:00
|
|
|
if (priv->opacity_parent)
|
2009-02-18 16:41:01 +00:00
|
|
|
return clutter_actor_get_paint_opacity (priv->opacity_parent);
|
|
|
|
|
|
|
|
parent = priv->parent_actor;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-01-04 19:56:01 +00:00
|
|
|
/* Factor in the actual actors opacity with parents */
|
2008-02-20 10:59:47 +00:00
|
|
|
if (G_LIKELY (parent))
|
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
guint8 opacity = clutter_actor_get_paint_opacity (parent);
|
2008-02-20 10:59:47 +00:00
|
|
|
|
|
|
|
if (opacity != 0xff)
|
|
|
|
return (opacity * priv->opacity) / 0xff;
|
|
|
|
}
|
|
|
|
|
|
|
|
return clutter_actor_get_opacity (self);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_opacity:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the opacity value of an actor, as set by
|
|
|
|
* clutter_actor_set_opacity().
|
|
|
|
*
|
|
|
|
* For retrieving the absolute opacity of the actor inside a paint
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
* virtual function, see clutter_actor_get_paint_opacity().
|
2008-02-20 10:59:47 +00:00
|
|
|
*
|
|
|
|
* Return value: the opacity of the actor
|
|
|
|
*/
|
|
|
|
guint8
|
|
|
|
clutter_actor_get_opacity (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0);
|
2006-06-13 13:17:45 +00:00
|
|
|
|
|
|
|
return self->priv->opacity;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_set_name:
|
|
|
|
* @self: A #ClutterActor
|
2006-07-06 18:55:46 +00:00
|
|
|
* @name: Textual tag to apply to actor
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* Sets the given name to @self. The name can be used to identify
|
|
|
|
* a #ClutterActor.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_name (ClutterActor *self,
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
const gchar *name)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
g_free (self->priv->name);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
self->priv->name = g_strdup (name);
|
2006-11-20 12:02:24 +00:00
|
|
|
|
|
|
|
g_object_notify (G_OBJECT (self), "name");
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_name:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2006-06-22 13:05:45 +00:00
|
|
|
* Retrieves the name of @self.
|
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Return value: the name of the actor, or %NULL. The returned string is
|
2008-02-15 14:39:25 +00:00
|
|
|
* owned by the actor and should not be modified or freed.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
G_CONST_RETURN gchar *
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_get_name (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), NULL);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
return self->priv->name;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2007-11-14 11:43:18 +00:00
|
|
|
* clutter_actor_get_gid:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2006-06-21 23:02:20 +00:00
|
|
|
* Retrieves the unique id for @self.
|
2007-10-12 08:17:00 +00:00
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Return value: Globally unique value for this object instance.
|
2007-11-14 11:43:18 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
guint32
|
2007-11-14 11:43:18 +00:00
|
|
|
clutter_actor_get_gid (ClutterActor *self)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
return self->priv->id;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_set_depth:
|
|
|
|
* @self: a #ClutterActor
|
2006-06-21 23:02:20 +00:00
|
|
|
* @depth: Z co-ord
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Sets the Z coordinate of @self to @depth.
|
2007-11-29 13:47:37 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The unit used by @depth is dependant on the perspective setup. See
|
|
|
|
* also clutter_stage_set_perspective().
|
2007-11-29 13:47:37 +00:00
|
|
|
*/
|
|
|
|
void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_set_depth (ClutterActor *self,
|
|
|
|
gfloat depth)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2006-12-12 20:20:04 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
2007-11-28 12:23:31 +00:00
|
|
|
|
2007-10-12 12:42:54 +00:00
|
|
|
if (priv->z != depth)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
/* Sets Z value - XXX 2.0: should we invert? */
|
2007-10-12 12:42:54 +00:00
|
|
|
priv->z = depth;
|
|
|
|
|
|
|
|
if (priv->parent_actor && CLUTTER_IS_CONTAINER (priv->parent_actor))
|
|
|
|
{
|
|
|
|
ClutterContainer *parent;
|
|
|
|
|
|
|
|
/* We need to resort the container stacking order as to
|
|
|
|
* correctly render alpha values.
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* FIXME: This is sub-optimal. maybe queue the the sort
|
2007-10-12 12:42:54 +00:00
|
|
|
* before stacking
|
|
|
|
*/
|
|
|
|
parent = CLUTTER_CONTAINER (priv->parent_actor);
|
|
|
|
clutter_container_sort_depth_order (parent);
|
|
|
|
}
|
2006-12-12 20:20:04 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_queue_redraw (self);
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-10-12 12:42:54 +00:00
|
|
|
g_object_notify (G_OBJECT (self), "depth");
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_depth:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the depth of @self.
|
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Return value: the depth of the actor
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_get_depth (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), -1);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
return self->priv->z;
|
|
|
|
}
|
|
|
|
|
2008-04-23 12:53:25 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_set_rotation:
|
2008-04-23 12:53:25 +00:00
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @axis: the axis of rotation
|
|
|
|
* @angle: the angle of rotation
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* @x: X coordinate of the rotation center
|
|
|
|
* @y: Y coordinate of the rotation center
|
|
|
|
* @z: Z coordinate of the rotation center
|
2008-04-23 12:53:25 +00:00
|
|
|
*
|
|
|
|
* Sets the rotation angle of @self around the given axis.
|
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* The rotation center coordinates used depend on the value of @axis:
|
|
|
|
* <itemizedlist>
|
|
|
|
* <listitem><para>%CLUTTER_X_AXIS requires @y and @z</para></listitem>
|
|
|
|
* <listitem><para>%CLUTTER_Y_AXIS requires @x and @z</para></listitem>
|
|
|
|
* <listitem><para>%CLUTTER_Z_AXIS requires @x and @y</para></listitem>
|
|
|
|
* </itemizedlist>
|
|
|
|
*
|
|
|
|
* The rotation coordinates are relative to the anchor point of the
|
|
|
|
* actor, set using clutter_actor_set_anchor_point(). If no anchor
|
|
|
|
* point is set, the upper left corner is assumed as the origin.
|
2008-04-23 12:53:25 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_set_rotation (ClutterActor *self,
|
|
|
|
ClutterRotateAxis axis,
|
|
|
|
gdouble angle,
|
|
|
|
gfloat x,
|
|
|
|
gfloat y,
|
|
|
|
gfloat z)
|
2008-04-23 12:53:25 +00:00
|
|
|
{
|
2009-01-23 15:55:41 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
2008-04-23 12:53:25 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
priv = self->priv;
|
2008-04-23 12:53:25 +00:00
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_set_rotation_internal (self, axis, angle);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
switch (axis)
|
|
|
|
{
|
|
|
|
case CLUTTER_X_AXIS:
|
|
|
|
clutter_anchor_coord_set_units (&priv->rx_center, x, y, z);
|
|
|
|
g_object_notify (G_OBJECT (self), "rotation-center-x");
|
|
|
|
break;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
case CLUTTER_Y_AXIS:
|
|
|
|
clutter_anchor_coord_set_units (&priv->ry_center, x, y, z);
|
|
|
|
g_object_notify (G_OBJECT (self), "rotation-center-y");
|
|
|
|
break;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
case CLUTTER_Z_AXIS:
|
|
|
|
if (priv->rz_center.is_fractional)
|
|
|
|
g_object_notify (G_OBJECT (self), "rotation-center-z-gravity");
|
|
|
|
clutter_anchor_coord_set_units (&priv->rz_center, x, y, z);
|
|
|
|
g_object_notify (G_OBJECT (self), "rotation-center-z");
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2007-06-26 11:35:33 +00:00
|
|
|
}
|
|
|
|
|
2009-01-23 17:25:43 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_z_rotation_from_gravity:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @angle: the angle of rotation
|
|
|
|
* @gravity: the center point of the rotation
|
|
|
|
*
|
|
|
|
* Sets the rotation angle of @self around the Z axis using the center
|
|
|
|
* point specified as a compass point. For example to rotate such that
|
|
|
|
* the center of the actor remains static you can use
|
|
|
|
* %CLUTTER_GRAVITY_CENTER. If the actor changes size the center point
|
|
|
|
* will move accordingly.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
2009-01-23 15:55:41 +00:00
|
|
|
void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_set_z_rotation_from_gravity (ClutterActor *self,
|
|
|
|
gdouble angle,
|
|
|
|
ClutterGravity gravity)
|
2009-01-23 15:55:41 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
if (gravity == CLUTTER_GRAVITY_NONE)
|
|
|
|
clutter_actor_set_rotation (self, CLUTTER_Z_AXIS, angle, 0, 0, 0);
|
|
|
|
else
|
|
|
|
{
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
|
|
|
clutter_actor_set_rotation_internal (self, CLUTTER_Z_AXIS, angle);
|
|
|
|
|
|
|
|
clutter_anchor_coord_set_gravity (&priv->rz_center, gravity);
|
|
|
|
g_object_notify (G_OBJECT (self), "rotation-center-z-gravity");
|
|
|
|
g_object_notify (G_OBJECT (self), "rotation-center-z");
|
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
|
|
|
}
|
2007-06-26 11:35:33 +00:00
|
|
|
}
|
|
|
|
|
2008-04-23 12:53:25 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_get_rotation:
|
2008-04-23 12:53:25 +00:00
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @axis: the axis of rotation
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* @x: (out): return value for the X coordinate of the center of rotation
|
|
|
|
* @y: (out): return value for the Y coordinate of the center of rotation
|
|
|
|
* @z: (out): return value for the Z coordinate of the center of rotation
|
2008-04-23 12:53:25 +00:00
|
|
|
*
|
|
|
|
* Retrieves the angle and center of rotation on the given axis,
|
|
|
|
* set using clutter_actor_set_rotation().
|
|
|
|
*
|
|
|
|
* Return value: the angle of rotation
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
gdouble
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_get_rotation (ClutterActor *self,
|
|
|
|
ClutterRotateAxis axis,
|
|
|
|
gfloat *x,
|
|
|
|
gfloat *y,
|
|
|
|
gfloat *z)
|
2008-04-23 12:53:25 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
gdouble retval = 0;
|
2009-01-23 15:55:41 +00:00
|
|
|
AnchorCoord *anchor_coord = NULL;
|
2008-04-23 12:53:25 +00:00
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0);
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
switch (axis)
|
|
|
|
{
|
|
|
|
case CLUTTER_X_AXIS:
|
2009-01-23 15:55:41 +00:00
|
|
|
anchor_coord = &priv->rx_center;
|
2007-11-18 15:36:04 +00:00
|
|
|
retval = priv->rxang;
|
|
|
|
break;
|
2007-08-01 13:11:43 +00:00
|
|
|
|
2007-11-18 15:36:04 +00:00
|
|
|
case CLUTTER_Y_AXIS:
|
2009-01-23 15:55:41 +00:00
|
|
|
anchor_coord = &priv->ry_center;
|
2007-11-18 15:36:04 +00:00
|
|
|
retval = priv->ryang;
|
|
|
|
break;
|
2007-08-01 13:11:43 +00:00
|
|
|
|
2007-11-18 15:36:04 +00:00
|
|
|
case CLUTTER_Z_AXIS:
|
2009-01-23 15:55:41 +00:00
|
|
|
anchor_coord = &priv->rz_center;
|
2007-11-18 15:36:04 +00:00
|
|
|
retval = priv->rzang;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
clutter_anchor_coord_get_units (self, anchor_coord, x, y, z);
|
|
|
|
|
2007-11-18 15:36:04 +00:00
|
|
|
return retval;
|
2007-06-26 11:35:33 +00:00
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2009-01-23 17:25:43 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_z_rotation_gravity:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the center for the rotation around the Z axis as a
|
|
|
|
* compass direction. If the center was specified in pixels or units
|
|
|
|
* this will return %CLUTTER_GRAVITY_NONE.
|
|
|
|
*
|
|
|
|
* Return value: the Z rotation center
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
2009-01-23 15:55:41 +00:00
|
|
|
ClutterGravity
|
|
|
|
clutter_actor_get_z_rotation_gravity (ClutterActor *self)
|
2007-08-01 13:11:43 +00:00
|
|
|
{
|
2007-11-18 15:36:04 +00:00
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0.0);
|
2007-08-01 13:11:43 +00:00
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
return clutter_anchor_coord_get_gravity (&self->priv->rz_center);
|
2007-08-01 13:11:43 +00:00
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_set_clip:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* @xoff: X offset of the clip rectangle
|
|
|
|
* @yoff: Y offset of the clip rectangle
|
|
|
|
* @width: Width of the clip rectangle
|
|
|
|
* @height: Height of the clip rectangle
|
2007-12-24 14:21:19 +00:00
|
|
|
*
|
|
|
|
* Sets clip area for @self. The clip area is always computed from the
|
|
|
|
* upper left corner of the actor, even if the anchor point is set
|
|
|
|
* otherwise.
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_set_clip (ClutterActor *self,
|
|
|
|
gfloat xoff,
|
|
|
|
gfloat yoff,
|
|
|
|
gfloat width,
|
|
|
|
gfloat height)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2007-12-24 12:53:04 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-12-24 12:53:04 +00:00
|
|
|
priv = self->priv;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2009-01-26 16:22:39 +00:00
|
|
|
if (priv->has_clip &&
|
|
|
|
priv->clip[0] == xoff &&
|
|
|
|
priv->clip[1] == yoff &&
|
|
|
|
priv->clip[2] == width &&
|
|
|
|
priv->clip[3] == height)
|
|
|
|
return;
|
|
|
|
|
2007-12-24 14:21:19 +00:00
|
|
|
priv->clip[0] = xoff;
|
|
|
|
priv->clip[1] = yoff;
|
|
|
|
priv->clip[2] = width;
|
|
|
|
priv->clip[3] = height;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2007-12-24 12:53:04 +00:00
|
|
|
priv->has_clip = TRUE;
|
2006-07-06 18:12:19 +00:00
|
|
|
|
2008-07-25 11:29:12 +00:00
|
|
|
clutter_actor_queue_redraw (self);
|
|
|
|
|
2006-07-06 18:12:19 +00:00
|
|
|
g_object_notify (G_OBJECT (self), "has-clip");
|
|
|
|
g_object_notify (G_OBJECT (self), "clip");
|
2007-10-12 08:17:00 +00:00
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_remove_clip
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Removes clip area from @self.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_remove_clip (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2009-01-26 16:22:39 +00:00
|
|
|
if (!self->priv->has_clip)
|
|
|
|
return;
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
self->priv->has_clip = FALSE;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-07-25 11:29:12 +00:00
|
|
|
clutter_actor_queue_redraw (self);
|
|
|
|
|
2006-07-06 18:12:19 +00:00
|
|
|
g_object_notify (G_OBJECT (self), "has-clip");
|
2007-10-12 08:17:00 +00:00
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2006-07-06 18:12:19 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_has_clip:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Determines whether the actor has a clip area set or not.
|
2007-10-12 08:17:00 +00:00
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Return value: %TRUE if the actor has a clip area set.
|
2006-07-06 18:55:46 +00:00
|
|
|
*
|
|
|
|
* Since: 0.1.1
|
2006-07-06 18:12:19 +00:00
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_has_clip (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), FALSE);
|
|
|
|
|
|
|
|
return self->priv->has_clip;
|
|
|
|
}
|
|
|
|
|
2007-11-15 17:35:47 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_get_clip:
|
2007-11-15 17:35:47 +00:00
|
|
|
* @self: a #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @xoff: (out) (allow-none): return location for the X offset of
|
|
|
|
* the clip rectangle, or %NULL
|
|
|
|
* @yoff: (out) (allow-none): return location for the Y offset of
|
|
|
|
* the clip rectangle, or %NULL
|
|
|
|
* @width: (out) (allow-none): return location for the width of
|
|
|
|
* the clip rectangle, or %NULL
|
|
|
|
* @height: (out) (allow-none): return location for the height of
|
|
|
|
* the clip rectangle, or %NULL
|
2007-11-15 17:35:47 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Gets the clip area for @self, if any is set
|
2007-11-15 17:35:47 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_get_clip (ClutterActor *self,
|
|
|
|
gfloat *xoff,
|
|
|
|
gfloat *yoff,
|
|
|
|
gfloat *width,
|
|
|
|
gfloat *height)
|
2007-11-15 17:35:47 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (!priv->has_clip)
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (xoff)
|
2007-12-24 14:21:19 +00:00
|
|
|
*xoff = priv->clip[0];
|
|
|
|
|
|
|
|
if (yoff)
|
|
|
|
*yoff = priv->clip[1];
|
|
|
|
|
|
|
|
if (width)
|
|
|
|
*width = priv->clip[2];
|
|
|
|
|
|
|
|
if (height)
|
|
|
|
*height = priv->clip[3];
|
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_parent:
|
|
|
|
* @self: A #ClutterActor
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
* @parent: A new #ClutterActor parent
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
* Sets the parent of @self to @parent. The opposite function is
|
|
|
|
* clutter_actor_unparent().
|
2007-10-12 08:17:00 +00:00
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* This function should not be used by applications, but by custom
|
2008-06-17 09:33:18 +00:00
|
|
|
* container actor subclasses.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_parent (ClutterActor *self,
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
ClutterActor *parent)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2009-11-30 16:14:09 +00:00
|
|
|
ClutterTextDirection text_dir;
|
actor: Add internal child flag
ClutterActor checks, when destroying and reparenting, if the parent
actor implements the Container interface, and automatically calls the
remove() method to perform a clean removal.
Actors implementing Container, though, might have internal children;
that is, children that are not added through the Container API. It is
already possible to iterate through them using the Container API to
avoid breaking invariants - but calling clutter_actor_destroy() on
these children (even from the Container implementation, and thus outside
of Clutter's control) will either lead to leaks or to segmentation
faults.
Clutter needs a way to distinguish a clutter_actor_set_parent() done on
an internal child from one done on a "public" child; for this reason, a
push/pop pair of functions should be available to Actor implementations
to mark the section where they wish to add internal children:
➔ clutter_actor_push_internal ();
...
clutter_actor_set_parent (child1, parent);
clutter_actor_set_parent (child2, parent);
...
➔ clutter_actor_pop_internal ();
The set_parent() call will automatically set the newly added
INTERNAL_CHILD private flag on each child, and both
clutter_actor_destroy() and clutter_actor_unparent() will check for the
flag before deciding whether to call the Container's remove method.
2009-12-18 23:20:04 +00:00
|
|
|
ClutterMainContext *ctx;
|
2007-10-08 16:18:33 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (parent));
|
|
|
|
g_return_if_fail (self != parent);
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->parent_actor != NULL)
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
{
|
|
|
|
g_warning ("Cannot set a parent on an actor which has a parent.\n"
|
|
|
|
"You must use clutter_actor_unparent() first.\n");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL)
|
|
|
|
{
|
|
|
|
g_warning ("Cannot set a parent on a toplevel actor\n");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2009-10-23 09:08:02 +00:00
|
|
|
if (CLUTTER_PRIVATE_FLAGS (parent) & CLUTTER_ACTOR_IN_DESTRUCTION)
|
|
|
|
{
|
|
|
|
g_warning ("Cannot set a parent currently being destroyed");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
g_object_ref_sink (self);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
priv->parent_actor = parent;
|
2008-06-17 12:39:37 +00:00
|
|
|
|
actor: Add internal child flag
ClutterActor checks, when destroying and reparenting, if the parent
actor implements the Container interface, and automatically calls the
remove() method to perform a clean removal.
Actors implementing Container, though, might have internal children;
that is, children that are not added through the Container API. It is
already possible to iterate through them using the Container API to
avoid breaking invariants - but calling clutter_actor_destroy() on
these children (even from the Container implementation, and thus outside
of Clutter's control) will either lead to leaks or to segmentation
faults.
Clutter needs a way to distinguish a clutter_actor_set_parent() done on
an internal child from one done on a "public" child; for this reason, a
push/pop pair of functions should be available to Actor implementations
to mark the section where they wish to add internal children:
➔ clutter_actor_push_internal ();
...
clutter_actor_set_parent (child1, parent);
clutter_actor_set_parent (child2, parent);
...
➔ clutter_actor_pop_internal ();
The set_parent() call will automatically set the newly added
INTERNAL_CHILD private flag on each child, and both
clutter_actor_destroy() and clutter_actor_unparent() will check for the
flag before deciding whether to call the Container's remove method.
2009-12-18 23:20:04 +00:00
|
|
|
ctx = _clutter_context_get_default ();
|
|
|
|
|
|
|
|
/* if push_internal() has been called then we automatically set
|
|
|
|
* the flag on the actor
|
|
|
|
*/
|
|
|
|
if (ctx->internal_child)
|
|
|
|
CLUTTER_SET_PRIVATE_FLAGS (self, CLUTTER_ACTOR_INTERNAL_CHILD);
|
|
|
|
|
2008-06-17 12:39:37 +00:00
|
|
|
/* clutter_actor_reparent() will emit ::parent-set for us */
|
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_REPARENT))
|
|
|
|
g_signal_emit (self, actor_signals[PARENT_SET], 0, NULL);
|
2006-11-20 12:02:24 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* If parent is mapped or realized, we need to also be mapped or
|
|
|
|
* realized once we're inside the parent.
|
2008-06-17 09:33:18 +00:00
|
|
|
*/
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_CHECK);
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2009-11-30 16:14:09 +00:00
|
|
|
/* propagate the parent's text direction to the child */
|
|
|
|
text_dir = clutter_actor_get_text_direction (parent);
|
|
|
|
clutter_actor_set_text_direction (self, text_dir);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->show_on_set_parent)
|
2008-05-23 09:42:32 +00:00
|
|
|
clutter_actor_show (self);
|
2008-06-16 09:27:44 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
if (CLUTTER_ACTOR_IS_MAPPED (self))
|
2009-11-30 16:14:09 +00:00
|
|
|
clutter_actor_queue_redraw (self);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
|
2008-06-17 09:33:18 +00:00
|
|
|
/* maintain the invariant that if an actor needs layout,
|
|
|
|
* its parents do as well
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*/
|
|
|
|
if (priv->needs_width_request ||
|
|
|
|
priv->needs_height_request ||
|
|
|
|
priv->needs_allocation)
|
|
|
|
{
|
2009-02-05 11:04:34 +00:00
|
|
|
/* we work around the short-circuiting we do
|
|
|
|
* in clutter_actor_queue_relayout() since we
|
|
|
|
* want to force a relayout
|
|
|
|
*/
|
|
|
|
priv->needs_width_request = TRUE;
|
|
|
|
priv->needs_height_request = TRUE;
|
|
|
|
priv->needs_allocation = TRUE;
|
|
|
|
|
|
|
|
clutter_actor_queue_relayout (priv->parent_actor);
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_parent:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2006-06-21 23:02:20 +00:00
|
|
|
* Retrieves the parent of @self.
|
|
|
|
*
|
2009-07-02 13:21:16 +00:00
|
|
|
* Return Value: (transfer none): The #ClutterActor parent, or %NULL
|
|
|
|
* if no parent is set
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
ClutterActor *
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_get_parent (ClutterActor *self)
|
|
|
|
{
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), NULL);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
return self->priv->parent_actor;
|
|
|
|
}
|
|
|
|
|
2008-11-07 16:31:56 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_paint_visibility:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the 'paint' visibility of an actor recursively checking for non
|
|
|
|
* visible parents.
|
|
|
|
*
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* This is by definition the same as CLUTTER_ACTOR_IS_MAPPED().
|
|
|
|
*
|
2008-11-07 16:31:56 +00:00
|
|
|
* Return Value: TRUE if the actor is visibile and will be painted.
|
|
|
|
*
|
|
|
|
* Since: 0.8.4
|
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_get_paint_visibility (ClutterActor *actor)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (actor), FALSE);
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
return CLUTTER_ACTOR_IS_MAPPED (actor);
|
2008-11-07 16:31:56 +00:00
|
|
|
}
|
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_unparent:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
2008-06-17 09:33:18 +00:00
|
|
|
* Removes the parent of @self.
|
|
|
|
*
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
* This function should not be used in applications. It should be called by
|
2007-08-13 17:01:07 +00:00
|
|
|
* implementations of container actors, to dissociate a child from the
|
|
|
|
* container.
|
2006-07-06 18:55:46 +00:00
|
|
|
*
|
|
|
|
* Since: 0.1.1
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_unparent (ClutterActor *self)
|
|
|
|
{
|
2008-06-17 09:33:18 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2006-11-20 12:02:24 +00:00
|
|
|
ClutterActor *old_parent;
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
gboolean was_mapped;
|
2006-11-20 12:02:24 +00:00
|
|
|
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2008-06-17 09:33:18 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->parent_actor == NULL)
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
return;
|
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
was_mapped = CLUTTER_ACTOR_IS_MAPPED (self);
|
2008-06-17 09:33:18 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* we need to unrealize *before* we set parent_actor to NULL,
|
|
|
|
* because in an unrealize method actors are dissociating from the
|
|
|
|
* stage, which means they need to be able to
|
|
|
|
* clutter_actor_get_stage(). This should unmap and unrealize,
|
|
|
|
* unless we're reparenting.
|
2008-06-17 09:33:18 +00:00
|
|
|
*/
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_MAKE_UNREALIZED);
|
2008-06-17 09:33:18 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
old_parent = priv->parent_actor;
|
|
|
|
priv->parent_actor = NULL;
|
2008-06-16 13:40:39 +00:00
|
|
|
|
2008-06-17 12:39:37 +00:00
|
|
|
/* clutter_actor_reparent() will emit ::parent-set for us */
|
|
|
|
if (!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IN_REPARENT))
|
|
|
|
g_signal_emit (self, actor_signals[PARENT_SET], 0, old_parent);
|
2006-11-20 12:02:24 +00:00
|
|
|
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
/* Queue a redraw on old_parent only if we were painted in the first
|
|
|
|
* place. Will be no-op if old parent is not shown.
|
|
|
|
*/
|
|
|
|
if (was_mapped && !CLUTTER_ACTOR_IS_MAPPED (self))
|
2008-06-17 13:49:40 +00:00
|
|
|
clutter_actor_queue_redraw (old_parent);
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
/* remove the reference we acquired in clutter_actor_set_parent() */
|
2006-07-06 Emmanuele Bassi <ebassi@openedhand.com>
Big rework of the actor management semantics: now ClutterActor
objects behave like GtkObjects - that is they have an initial
"floating" reference that gets "sunk" when they are added to
a ClutterGroup. This makes a group responsible of de-allocating
each actor inside it, so you just have to destroy the group to
get every child actor destroyed. Also, now you can do:
clutter_group_add (group, clutter_video_texture_new ());
without having to care about reference counting and explicit
unreffing.
* clutter/clutter-private.h: Add private flags setter and
getter macros.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c: Clean up; inherit from GInitiallyUnowned;
add a "visible" property; add the "destroy", "show" and "hide"
signals to ClutterActorClass.
(clutter_actor_show), (clutter_actor_hide): Refactor a bit; emit
the "show" and "hide" signals.
(clutter_actor_set_property), (clutter_actor_get_property),
(clutter_actor_class_init): Implement the "visible" property; add
signals.
(clutter_actor_finalize): Do not leak the actor's name, if it is
set.
(clutter_actor_dispose): Emit the "destroy" signal here.
(clutter_actor_init): Sink the initial floating flag if needed.
(clutter_actor_destroy): Add a function to explicitely destroy
a ClutterActor.
(clutter_actor_set_parent), (clutter_actor_get_parent),
(clutter_actor_unparent): Make set_parent require a valid parent;
add unparent; check on get_parent; ref_sink the actor when
setting its parent and unref it when unsetting it. Probably we'll
need a function that does reparenting as unparent+set_parent in
a single shot.
* clutter/clutter-group.h:
* clutter/clutter-group.c (clutter_group_dispose),
(clutter_group_finalize), (clutter_group_add),
(clutter_group_remove): Make the group destroy its children when
disposing it; clean up, and use the newly-available
clutter_actor_unparent().
* clutter/clutter-stage.h:
* clutter/clutter-stage.c (clutter_stage_init): ClutterStage is
a top-level actor; clean up.
* clutter/clutter-video-texture.h:
* clutter/clutter-video-texture.c: Clean up.
* examples/super-oh.c:
* examples/test.c:
* examples/video-player.c:
* examples/test-text.c:
* examples/video-cube.c: Remove the g_object_unref() call, as the
ClutterStage object is destroyed on clutter_main_quit().
2006-07-06 17:52:57 +00:00
|
|
|
g_object_unref (self);
|
|
|
|
}
|
|
|
|
|
2006-11-16 00:24:23 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_reparent:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @new_parent: the new #ClutterActor parent
|
|
|
|
*
|
|
|
|
* This function resets the parent actor of @self. It is
|
2008-02-15 15:18:02 +00:00
|
|
|
* logically equivalent to calling clutter_actor_unparent()
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
* and clutter_actor_set_parent(), but more efficiently
|
|
|
|
* implemented, ensures the child is not finalized
|
|
|
|
* when unparented, and emits the parent-set signal only
|
|
|
|
* one time.
|
2006-11-16 00:24:23 +00:00
|
|
|
*
|
2006-11-20 12:02:24 +00:00
|
|
|
* Since: 0.2
|
2006-11-16 00:24:23 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_reparent (ClutterActor *self,
|
|
|
|
ClutterActor *new_parent)
|
|
|
|
{
|
2007-06-07 14:41:35 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
2006-11-16 00:24:23 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (new_parent));
|
|
|
|
g_return_if_fail (self != new_parent);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2006-11-16 00:24:23 +00:00
|
|
|
if (CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_IS_TOPLEVEL)
|
|
|
|
{
|
2009-10-23 09:08:02 +00:00
|
|
|
g_warning ("Cannot set a parent on a toplevel actor");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (CLUTTER_PRIVATE_FLAGS (new_parent) & CLUTTER_ACTOR_IN_DESTRUCTION)
|
|
|
|
{
|
|
|
|
g_warning ("Cannot set a parent currently being destroyed");
|
2006-11-16 00:24:23 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2007-06-07 14:41:35 +00:00
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->parent_actor != new_parent)
|
2006-11-16 00:24:23 +00:00
|
|
|
{
|
2006-11-20 12:02:24 +00:00
|
|
|
ClutterActor *old_parent;
|
|
|
|
|
2008-06-17 12:39:37 +00:00
|
|
|
CLUTTER_SET_PRIVATE_FLAGS (self, CLUTTER_ACTOR_IN_REPARENT);
|
2006-11-16 00:24:23 +00:00
|
|
|
|
2007-06-07 14:41:35 +00:00
|
|
|
old_parent = priv->parent_actor;
|
2006-11-20 12:02:24 +00:00
|
|
|
|
2006-11-16 00:24:23 +00:00
|
|
|
g_object_ref (self);
|
2007-06-07 14:41:35 +00:00
|
|
|
|
actor: Add internal child flag
ClutterActor checks, when destroying and reparenting, if the parent
actor implements the Container interface, and automatically calls the
remove() method to perform a clean removal.
Actors implementing Container, though, might have internal children;
that is, children that are not added through the Container API. It is
already possible to iterate through them using the Container API to
avoid breaking invariants - but calling clutter_actor_destroy() on
these children (even from the Container implementation, and thus outside
of Clutter's control) will either lead to leaks or to segmentation
faults.
Clutter needs a way to distinguish a clutter_actor_set_parent() done on
an internal child from one done on a "public" child; for this reason, a
push/pop pair of functions should be available to Actor implementations
to mark the section where they wish to add internal children:
➔ clutter_actor_push_internal ();
...
clutter_actor_set_parent (child1, parent);
clutter_actor_set_parent (child2, parent);
...
➔ clutter_actor_pop_internal ();
The set_parent() call will automatically set the newly added
INTERNAL_CHILD private flag on each child, and both
clutter_actor_destroy() and clutter_actor_unparent() will check for the
flag before deciding whether to call the Container's remove method.
2009-12-18 23:20:04 +00:00
|
|
|
/* go through the Container implementation if this is a regular
|
|
|
|
* child and not an internal one
|
|
|
|
*/
|
|
|
|
if (CLUTTER_IS_CONTAINER (priv->parent_actor) &&
|
|
|
|
!(CLUTTER_PRIVATE_FLAGS (self) & CLUTTER_ACTOR_INTERNAL_CHILD))
|
2008-06-17 12:39:37 +00:00
|
|
|
{
|
|
|
|
ClutterContainer *parent = CLUTTER_CONTAINER (priv->parent_actor);
|
actor: Add internal child flag
ClutterActor checks, when destroying and reparenting, if the parent
actor implements the Container interface, and automatically calls the
remove() method to perform a clean removal.
Actors implementing Container, though, might have internal children;
that is, children that are not added through the Container API. It is
already possible to iterate through them using the Container API to
avoid breaking invariants - but calling clutter_actor_destroy() on
these children (even from the Container implementation, and thus outside
of Clutter's control) will either lead to leaks or to segmentation
faults.
Clutter needs a way to distinguish a clutter_actor_set_parent() done on
an internal child from one done on a "public" child; for this reason, a
push/pop pair of functions should be available to Actor implementations
to mark the section where they wish to add internal children:
➔ clutter_actor_push_internal ();
...
clutter_actor_set_parent (child1, parent);
clutter_actor_set_parent (child2, parent);
...
➔ clutter_actor_pop_internal ();
The set_parent() call will automatically set the newly added
INTERNAL_CHILD private flag on each child, and both
clutter_actor_destroy() and clutter_actor_unparent() will check for the
flag before deciding whether to call the Container's remove method.
2009-12-18 23:20:04 +00:00
|
|
|
|
|
|
|
/* this will have to call unparent() */
|
2008-06-17 12:39:37 +00:00
|
|
|
clutter_container_remove_actor (parent, self);
|
|
|
|
}
|
2007-10-17 16:43:46 +00:00
|
|
|
else
|
2008-06-17 12:39:37 +00:00
|
|
|
clutter_actor_unparent (self);
|
2007-10-17 16:43:46 +00:00
|
|
|
|
2008-06-17 13:49:40 +00:00
|
|
|
/* Note, will call parent() */
|
actor: Add internal child flag
ClutterActor checks, when destroying and reparenting, if the parent
actor implements the Container interface, and automatically calls the
remove() method to perform a clean removal.
Actors implementing Container, though, might have internal children;
that is, children that are not added through the Container API. It is
already possible to iterate through them using the Container API to
avoid breaking invariants - but calling clutter_actor_destroy() on
these children (even from the Container implementation, and thus outside
of Clutter's control) will either lead to leaks or to segmentation
faults.
Clutter needs a way to distinguish a clutter_actor_set_parent() done on
an internal child from one done on a "public" child; for this reason, a
push/pop pair of functions should be available to Actor implementations
to mark the section where they wish to add internal children:
➔ clutter_actor_push_internal ();
...
clutter_actor_set_parent (child1, parent);
clutter_actor_set_parent (child2, parent);
...
➔ clutter_actor_pop_internal ();
The set_parent() call will automatically set the newly added
INTERNAL_CHILD private flag on each child, and both
clutter_actor_destroy() and clutter_actor_unparent() will check for the
flag before deciding whether to call the Container's remove method.
2009-12-18 23:20:04 +00:00
|
|
|
if (CLUTTER_IS_CONTAINER (new_parent))
|
2007-10-17 16:43:46 +00:00
|
|
|
clutter_container_add_actor (CLUTTER_CONTAINER (new_parent), self);
|
|
|
|
else
|
2008-06-17 12:39:37 +00:00
|
|
|
clutter_actor_set_parent (self, new_parent);
|
|
|
|
|
|
|
|
/* we emit the ::parent-set signal once */
|
|
|
|
g_signal_emit (self, actor_signals[PARENT_SET], 0, old_parent);
|
2007-06-07 14:41:35 +00:00
|
|
|
|
2006-11-16 00:24:23 +00:00
|
|
|
g_object_unref (self);
|
|
|
|
|
2008-06-17 12:39:37 +00:00
|
|
|
CLUTTER_UNSET_PRIVATE_FLAGS (self, CLUTTER_ACTOR_IN_REPARENT);
|
Enforce invariants on mapped, realized, visibility states
Bug 1138 - No trackable "mapped" state
* Add a VISIBLE flag tracking application programmer's
expected showing-state for the actor, allowing us to
always ensure we keep what the app wants while tracking
internal implementation state separately.
* Make MAPPED reflect whether the actor will be painted;
add notification on a ClutterActor::mapped property.
Keep MAPPED state updated as the actor is shown,
ancestors are shown, actor is reparented, etc.
* Require a stage and realized parents to realize; this means
at realization time the correct window system and GL resources
are known. But unparented actors can no longer be realized.
* Allow children to be unrealized even if parent is realized.
Otherwise in effect either all actors or no actors are realized,
i.e. it becomes a stage-global flag.
* Allow clutter_actor_realize() to "fail" if not inside a toplevel
* Rework clutter_actor_unrealize() so internally we have
a flavor that does not mess with visibility flag
* Add _clutter_actor_rerealize() to encapsulate a somewhat
tricky operation we were doing in a couple of places
* Do not realize/unrealize children in ClutterGroup,
ClutterActor already does it
* Do not realize impl by hand in clutter_stage_show(),
since showing impl already does that
* Do not unrealize in various dispose() methods, since
ClutterActor dispose implementation already does it
and chaining up is mandatory
* ClutterTexture uses COGL while unrealizable (before it's
added to a stage). Previously this breakage was affecting
ClutterActor because we had to allow realize outside
a stage. Move the breakage to ClutterTexture, by making
ClutterTexture just use COGL while not realized.
* Unrealize before we set parent to NULL in clutter_actor_unparent().
This means unrealize() implementations can get to the stage.
Because actors need the stage in order to detach from stage.
* Update clutter-actor-invariants.txt to reflect latest changes
* Remove explicit hide/unrealize from ClutterActor::dispose since
unparent already forces those
Instead just assert that unparent() occurred and did the right thing.
* Check whether parent implements unrealize before chaining up
Needed because ClutterGroup no longer has to implement unrealize.
* Perform unrealize in the default handler for the signal.
This allows non-containers that have children to work properly,
and allows containers to override how it's done.
* Add map/unmap virtual methods and set MAPPED flag on self and
children in there. This allows subclasses to hook map/unmap.
These are not signals, because notify::mapped is better for
anything it's legitimate for a non-subclass to do.
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2009-04-02 13:16:43 +00:00
|
|
|
|
|
|
|
/* the IN_REPARENT flag suspends state updates */
|
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_CHECK);
|
2006-11-16 00:24:23 +00:00
|
|
|
}
|
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_raise:
|
|
|
|
* @self: A #ClutterActor
|
2009-02-17 00:25:20 +00:00
|
|
|
* @below: (allow-none): A #ClutterActor to raise above.
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2006-06-21 23:02:20 +00:00
|
|
|
* Puts @self above @below.
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
2006-06-13 13:17:45 +00:00
|
|
|
* Both actors must have the same parent.
|
2008-06-17 09:33:18 +00:00
|
|
|
*
|
|
|
|
* This function is the equivalent of clutter_container_raise_child().
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
2006-11-20 12:02:24 +00:00
|
|
|
clutter_actor_raise (ClutterActor *self,
|
|
|
|
ClutterActor *below)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2006-11-20 12:02:24 +00:00
|
|
|
ClutterActor *parent;
|
2006-06-13 13:17:45 +00:00
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
parent = clutter_actor_get_parent (self);
|
|
|
|
if (!parent)
|
|
|
|
{
|
2007-08-13 17:01:07 +00:00
|
|
|
g_warning ("Actor of type %s is not inside a container",
|
2006-11-20 12:02:24 +00:00
|
|
|
g_type_name (G_OBJECT_TYPE (self)));
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (below)
|
|
|
|
{
|
|
|
|
if (parent != clutter_actor_get_parent (below))
|
|
|
|
{
|
|
|
|
g_warning ("Actor of type %s is not in the same "
|
2007-08-13 17:01:07 +00:00
|
|
|
"container of actor of type %s",
|
2006-11-20 12:02:24 +00:00
|
|
|
g_type_name (G_OBJECT_TYPE (self)),
|
|
|
|
g_type_name (G_OBJECT_TYPE (below)));
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-08-14 08:01:33 +00:00
|
|
|
clutter_container_raise_child (CLUTTER_CONTAINER (parent), self, below);
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_lower:
|
|
|
|
* @self: A #ClutterActor
|
2009-02-17 00:25:20 +00:00
|
|
|
* @above: (allow-none): A #ClutterActor to lower below
|
2006-06-13 13:17:45 +00:00
|
|
|
*
|
2006-06-21 23:02:20 +00:00
|
|
|
* Puts @self below @above.
|
2008-06-17 09:33:18 +00:00
|
|
|
*
|
2006-06-13 13:17:45 +00:00
|
|
|
* Both actors must have the same parent.
|
2008-06-17 09:33:18 +00:00
|
|
|
*
|
|
|
|
* This function is the equivalent of clutter_container_lower_child().
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
2006-11-20 12:02:24 +00:00
|
|
|
clutter_actor_lower (ClutterActor *self,
|
|
|
|
ClutterActor *above)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2006-11-20 12:02:24 +00:00
|
|
|
ClutterActor *parent;
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR(self));
|
|
|
|
|
2006-11-20 12:02:24 +00:00
|
|
|
parent = clutter_actor_get_parent (self);
|
|
|
|
if (!parent)
|
2006-06-13 13:17:45 +00:00
|
|
|
{
|
2007-08-13 17:01:07 +00:00
|
|
|
g_warning ("Actor of type %s is not inside a container",
|
2006-11-20 12:02:24 +00:00
|
|
|
g_type_name (G_OBJECT_TYPE (self)));
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (above)
|
|
|
|
{
|
|
|
|
if (parent != clutter_actor_get_parent (above))
|
|
|
|
{
|
|
|
|
g_warning ("Actor of type %s is not in the same "
|
2007-08-13 17:01:07 +00:00
|
|
|
"container of actor of type %s",
|
2006-11-20 12:02:24 +00:00
|
|
|
g_type_name (G_OBJECT_TYPE (self)),
|
|
|
|
g_type_name (G_OBJECT_TYPE (above)));
|
|
|
|
return;
|
|
|
|
}
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
2007-08-14 08:01:33 +00:00
|
|
|
clutter_container_lower_child (CLUTTER_CONTAINER (parent), self, above);
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2006-06-22 13:05:45 +00:00
|
|
|
* clutter_actor_raise_top:
|
2006-06-13 13:17:45 +00:00
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
2006-06-21 23:02:20 +00:00
|
|
|
* Raises @self to the top.
|
2008-06-17 09:33:18 +00:00
|
|
|
*
|
|
|
|
* This function calls clutter_actor_raise() internally.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_raise_top (ClutterActor *self)
|
|
|
|
{
|
|
|
|
clutter_actor_raise (self, NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_lower_bottom:
|
|
|
|
* @self: A #ClutterActor
|
|
|
|
*
|
|
|
|
* Lowers @self to the bottom.
|
2008-06-17 09:33:18 +00:00
|
|
|
*
|
|
|
|
* This function calls clutter_actor_lower() internally.
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_lower_bottom (ClutterActor *self)
|
|
|
|
{
|
|
|
|
clutter_actor_lower (self, NULL);
|
|
|
|
}
|
|
|
|
|
2007-10-12 08:17:00 +00:00
|
|
|
/*
|
2007-08-13 20:48:01 +00:00
|
|
|
* Event handling
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_event:
|
|
|
|
* @actor: a #ClutterActor
|
|
|
|
* @event: a #ClutterEvent
|
2007-10-10 13:04:34 +00:00
|
|
|
* @capture: TRUE if event in in capture phase, FALSE otherwise.
|
2007-08-13 20:48:01 +00:00
|
|
|
*
|
|
|
|
* This function is used to emit an event on the main stage.
|
|
|
|
* You should rarely need to use this function, except for
|
|
|
|
* synthetising events.
|
|
|
|
*
|
2007-10-10 14:51:25 +00:00
|
|
|
* Return value: the return value from the signal emission: %TRUE
|
|
|
|
* if the actor handled the event, or %FALSE if the event was
|
|
|
|
* not handled
|
2007-08-13 20:48:01 +00:00
|
|
|
*
|
2007-08-15 19:50:31 +00:00
|
|
|
* Since: 0.6
|
2007-08-13 20:48:01 +00:00
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_event (ClutterActor *actor,
|
2007-10-10 13:04:34 +00:00
|
|
|
ClutterEvent *event,
|
|
|
|
gboolean capture)
|
2007-08-13 20:48:01 +00:00
|
|
|
{
|
2007-10-10 14:29:29 +00:00
|
|
|
gboolean retval = FALSE;
|
2007-08-13 20:48:01 +00:00
|
|
|
gint signal_num = -1;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (actor), FALSE);
|
|
|
|
g_return_val_if_fail (event != NULL, FALSE);
|
|
|
|
|
|
|
|
g_object_ref (actor);
|
2007-10-10 13:04:34 +00:00
|
|
|
|
|
|
|
if (capture)
|
|
|
|
{
|
2007-11-07 16:40:30 +00:00
|
|
|
g_signal_emit (actor, actor_signals[CAPTURED_EVENT], 0,
|
2007-10-10 14:29:29 +00:00
|
|
|
event,
|
|
|
|
&retval);
|
2007-10-10 13:04:34 +00:00
|
|
|
goto out;
|
|
|
|
}
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-10-02 14:03:36 +00:00
|
|
|
g_signal_emit (actor, actor_signals[EVENT], 0, event, &retval);
|
2007-08-13 20:48:01 +00:00
|
|
|
|
2007-10-02 14:03:36 +00:00
|
|
|
if (!retval)
|
2007-08-13 20:48:01 +00:00
|
|
|
{
|
2007-10-02 14:03:36 +00:00
|
|
|
switch (event->type)
|
2007-10-10 13:04:34 +00:00
|
|
|
{
|
|
|
|
case CLUTTER_NOTHING:
|
|
|
|
break;
|
|
|
|
case CLUTTER_BUTTON_PRESS:
|
|
|
|
signal_num = BUTTON_PRESS_EVENT;
|
|
|
|
break;
|
|
|
|
case CLUTTER_BUTTON_RELEASE:
|
|
|
|
signal_num = BUTTON_RELEASE_EVENT;
|
|
|
|
break;
|
|
|
|
case CLUTTER_SCROLL:
|
|
|
|
signal_num = SCROLL_EVENT;
|
|
|
|
break;
|
|
|
|
case CLUTTER_KEY_PRESS:
|
|
|
|
signal_num = KEY_PRESS_EVENT;
|
|
|
|
break;
|
|
|
|
case CLUTTER_KEY_RELEASE:
|
|
|
|
signal_num = KEY_RELEASE_EVENT;
|
|
|
|
break;
|
|
|
|
case CLUTTER_MOTION:
|
|
|
|
signal_num = MOTION_EVENT;
|
|
|
|
break;
|
|
|
|
case CLUTTER_ENTER:
|
|
|
|
signal_num = ENTER_EVENT;
|
|
|
|
break;
|
|
|
|
case CLUTTER_LEAVE:
|
|
|
|
signal_num = LEAVE_EVENT;
|
|
|
|
break;
|
|
|
|
case CLUTTER_DELETE:
|
|
|
|
case CLUTTER_DESTROY_NOTIFY:
|
|
|
|
case CLUTTER_CLIENT_MESSAGE:
|
|
|
|
default:
|
|
|
|
signal_num = -1;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2007-10-02 14:03:36 +00:00
|
|
|
if (signal_num != -1)
|
2007-10-12 08:17:00 +00:00
|
|
|
g_signal_emit (actor, actor_signals[signal_num], 0,
|
2007-10-10 13:04:34 +00:00
|
|
|
event, &retval);
|
2007-08-13 20:48:01 +00:00
|
|
|
}
|
|
|
|
|
2007-10-10 14:29:29 +00:00
|
|
|
out:
|
2007-08-13 20:48:01 +00:00
|
|
|
g_object_unref (actor);
|
|
|
|
|
2007-10-02 14:03:36 +00:00
|
|
|
return retval;
|
2007-08-13 20:48:01 +00:00
|
|
|
}
|
|
|
|
|
2007-10-10 16:02:54 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_reactive:
|
|
|
|
* @actor: a #ClutterActor
|
2007-11-15 17:35:47 +00:00
|
|
|
* @reactive: whether the actor should be reactive to events
|
2007-10-10 16:02:54 +00:00
|
|
|
*
|
|
|
|
* Sets @actor as reactive. Reactive actors will receive events.
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
2007-08-13 20:48:01 +00:00
|
|
|
void
|
2007-11-15 17:35:47 +00:00
|
|
|
clutter_actor_set_reactive (ClutterActor *actor,
|
|
|
|
gboolean reactive)
|
2007-08-13 20:48:01 +00:00
|
|
|
{
|
2007-08-15 19:50:31 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (actor));
|
|
|
|
|
2007-11-15 17:35:47 +00:00
|
|
|
if (reactive == CLUTTER_ACTOR_IS_REACTIVE (actor))
|
|
|
|
return;
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2007-11-15 17:35:47 +00:00
|
|
|
if (reactive)
|
|
|
|
CLUTTER_ACTOR_SET_FLAGS (actor, CLUTTER_ACTOR_REACTIVE);
|
|
|
|
else
|
|
|
|
CLUTTER_ACTOR_UNSET_FLAGS (actor, CLUTTER_ACTOR_REACTIVE);
|
2008-02-01 11:22:59 +00:00
|
|
|
|
|
|
|
g_object_notify (G_OBJECT (actor), "reactive");
|
2007-08-13 20:48:01 +00:00
|
|
|
}
|
|
|
|
|
2007-10-10 16:02:54 +00:00
|
|
|
/**
|
2007-11-15 17:35:47 +00:00
|
|
|
* clutter_actor_get_reactive:
|
2007-10-10 16:02:54 +00:00
|
|
|
* @actor: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Checks whether @actor is marked as reactive.
|
|
|
|
*
|
|
|
|
* Return value: %TRUE if the actor is reactive
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
2007-08-13 20:48:01 +00:00
|
|
|
gboolean
|
2007-11-15 17:35:47 +00:00
|
|
|
clutter_actor_get_reactive (ClutterActor *actor)
|
2007-08-13 20:48:01 +00:00
|
|
|
{
|
2007-08-15 19:50:31 +00:00
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (actor), FALSE);
|
2007-10-12 08:17:00 +00:00
|
|
|
|
2008-11-13 10:33:56 +00:00
|
|
|
return CLUTTER_ACTOR_IS_REACTIVE (actor) ? TRUE : FALSE;
|
2007-08-13 20:48:01 +00:00
|
|
|
}
|
|
|
|
|
2007-11-28 12:23:31 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_anchor_point:
|
2007-11-28 14:16:37 +00:00
|
|
|
* @self: a #ClutterActor
|
2009-02-26 19:42:19 +00:00
|
|
|
* @anchor_x: (out): return location for the X coordinate of the anchor point
|
|
|
|
* @anchor_y: (out): return location for the Y coordinate of the anchor point
|
2007-11-28 12:23:31 +00:00
|
|
|
*
|
2007-11-28 14:16:37 +00:00
|
|
|
* Gets the current anchor point of the @actor in pixels.
|
2007-11-28 12:23:31 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_anchor_point (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat *anchor_x,
|
|
|
|
gfloat *anchor_y)
|
2007-11-28 12:23:31 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_anchor_coord_get_units (self, &priv->anchor,
|
|
|
|
anchor_x,
|
|
|
|
anchor_y,
|
|
|
|
NULL);
|
2007-11-28 12:23:31 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_set_anchor_point:
|
2007-11-28 14:16:37 +00:00
|
|
|
* @self: a #ClutterActor
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* @anchor_x: X coordinate of the anchor point
|
|
|
|
* @anchor_y: Y coordinate of the anchor point
|
2007-11-28 12:23:31 +00:00
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Sets an anchor point for @self. The anchor point is a point in the
|
2007-11-28 14:16:37 +00:00
|
|
|
* coordinate space of an actor to which the actor position within its
|
|
|
|
* parent is relative; the default is (0, 0), i.e. the top-left corner
|
|
|
|
* of the actor.
|
2007-11-28 12:23:31 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_set_anchor_point (ClutterActor *self,
|
|
|
|
gfloat anchor_x,
|
|
|
|
gfloat anchor_y)
|
2007-11-28 12:23:31 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
2008-09-10 11:20:22 +00:00
|
|
|
gboolean changed = FALSE;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat old_anchor_x, old_anchor_y;
|
2007-11-28 12:23:31 +00:00
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
2008-03-06 04:37:00 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_anchor_coord_get_units (self, &priv->anchor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&old_anchor_x,
|
|
|
|
&old_anchor_y,
|
|
|
|
NULL);
|
2009-01-22 13:07:33 +00:00
|
|
|
|
|
|
|
if (priv->anchor.is_fractional)
|
|
|
|
g_object_notify (G_OBJECT (self), "anchor-gravity");
|
|
|
|
|
|
|
|
if (old_anchor_x != anchor_x)
|
2008-03-06 04:37:00 +00:00
|
|
|
{
|
|
|
|
g_object_notify (G_OBJECT (self), "anchor-x");
|
2008-09-10 11:20:22 +00:00
|
|
|
changed = TRUE;
|
2008-03-06 04:37:00 +00:00
|
|
|
}
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
if (old_anchor_y != anchor_y)
|
2008-03-06 04:37:00 +00:00
|
|
|
{
|
|
|
|
g_object_notify (G_OBJECT (self), "anchor-y");
|
2008-09-10 11:20:22 +00:00
|
|
|
changed = TRUE;
|
2008-03-06 04:37:00 +00:00
|
|
|
}
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_anchor_coord_set_units (&priv->anchor, anchor_x, anchor_y, 0);
|
|
|
|
|
2008-09-10 11:20:22 +00:00
|
|
|
if (changed && CLUTTER_ACTOR_IS_VISIBLE (self))
|
|
|
|
clutter_actor_queue_redraw (self);
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2007-11-28 12:23:31 +00:00
|
|
|
}
|
|
|
|
|
2009-01-23 17:25:43 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_anchor_point_gravity:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the anchor position expressed as a #ClutterGravity. If
|
|
|
|
* the anchor point was specified using pixels or units this will
|
|
|
|
* return %CLUTTER_GRAVITY_NONE.
|
|
|
|
*
|
2009-02-19 14:56:07 +00:00
|
|
|
* Return value: the #ClutterGravity used by the anchor point
|
|
|
|
*
|
2009-01-23 17:25:43 +00:00
|
|
|
* Since: 1.0
|
|
|
|
*/
|
2009-01-22 13:07:33 +00:00
|
|
|
ClutterGravity
|
|
|
|
clutter_actor_get_anchor_point_gravity (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), CLUTTER_GRAVITY_NONE);
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
return clutter_anchor_coord_get_gravity (&priv->anchor);
|
|
|
|
}
|
|
|
|
|
2008-01-30 12:13:26 +00:00
|
|
|
/**
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* clutter_actor_move_anchor_point:
|
2008-01-30 12:13:26 +00:00
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @anchor_x: X coordinate of the anchor point
|
|
|
|
* @anchor_y: Y coordinate of the anchor point
|
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Sets an anchor point for the actor, and adjusts the actor postion so that
|
2008-01-30 12:13:26 +00:00
|
|
|
* the relative position of the actor toward its parent remains the same.
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_move_anchor_point (ClutterActor *self,
|
|
|
|
gfloat anchor_x,
|
|
|
|
gfloat anchor_y)
|
2008-01-30 12:13:26 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat old_anchor_x, old_anchor_y;
|
2008-01-30 12:13:26 +00:00
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_anchor_coord_get_units (self, &priv->anchor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&old_anchor_x,
|
|
|
|
&old_anchor_y,
|
|
|
|
NULL);
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_actor_set_anchor_point (self, anchor_x, anchor_y);
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->position_set)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_move_by (self,
|
|
|
|
anchor_x - old_anchor_x,
|
|
|
|
anchor_y - old_anchor_y);
|
2009-01-22 13:07:33 +00:00
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2008-01-30 12:13:26 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_move_anchor_point_from_gravity:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @gravity: #ClutterGravity.
|
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Sets an anchor point on the actor based on the given gravity, adjusting the
|
|
|
|
* actor postion so that its relative position within its parent remains
|
2008-01-30 12:13:26 +00:00
|
|
|
* unchanged.
|
|
|
|
*
|
2009-01-23 17:25:43 +00:00
|
|
|
* Since version 1.0 the anchor point will be stored as a gravity so
|
|
|
|
* that if the actor changes size then the anchor point will move. For
|
|
|
|
* example, if you set the anchor point to %CLUTTER_GRAVITY_SOUTH_EAST
|
|
|
|
* and later double the size of the actor, the anchor point will move
|
|
|
|
* to the bottom right.
|
2008-09-17 10:40:32 +00:00
|
|
|
*
|
2008-01-30 12:13:26 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_move_anchor_point_from_gravity (ClutterActor *self,
|
|
|
|
ClutterGravity gravity)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat old_anchor_x, old_anchor_y, new_anchor_x, new_anchor_y;
|
2008-01-30 12:13:26 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
g_object_freeze_notify (G_OBJECT (self));
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_anchor_coord_get_units (self, &priv->anchor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&old_anchor_x,
|
|
|
|
&old_anchor_y,
|
|
|
|
NULL);
|
2008-01-30 12:13:26 +00:00
|
|
|
clutter_actor_set_anchor_point_from_gravity (self, gravity);
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_anchor_coord_get_units (self, &priv->anchor,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
&new_anchor_x,
|
|
|
|
&new_anchor_y,
|
|
|
|
NULL);
|
2008-01-30 12:13:26 +00:00
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
if (priv->position_set)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_move_by (self,
|
|
|
|
new_anchor_x - old_anchor_x,
|
|
|
|
new_anchor_y - old_anchor_y);
|
2009-01-22 13:07:33 +00:00
|
|
|
|
|
|
|
g_object_thaw_notify (G_OBJECT (self));
|
2008-01-30 12:13:26 +00:00
|
|
|
}
|
|
|
|
|
2007-11-28 12:23:31 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_anchor_point_from_gravity:
|
2007-11-28 14:16:37 +00:00
|
|
|
* @self: a #ClutterActor
|
2007-11-28 12:23:31 +00:00
|
|
|
* @gravity: #ClutterGravity.
|
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* Sets an anchor point on the actor, based on the given gravity (this is a
|
2007-11-28 14:16:37 +00:00
|
|
|
* convenience function wrapping clutter_actor_set_anchor_point()).
|
2007-11-28 12:23:31 +00:00
|
|
|
*
|
2009-01-23 17:25:43 +00:00
|
|
|
* Since version 1.0 the anchor point will be stored as a gravity so
|
|
|
|
* that if the actor changes size then the anchor point will move. For
|
|
|
|
* example, if you set the anchor point to %CLUTTER_GRAVITY_SOUTH_EAST
|
|
|
|
* and later double the size of the actor, the anchor point will move
|
|
|
|
* to the bottom right.
|
2008-09-17 10:40:32 +00:00
|
|
|
*
|
2007-11-28 12:23:31 +00:00
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
void
|
2007-11-28 14:16:37 +00:00
|
|
|
clutter_actor_set_anchor_point_from_gravity (ClutterActor *self,
|
|
|
|
ClutterGravity gravity)
|
2007-11-28 12:23:31 +00:00
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
if (gravity == CLUTTER_GRAVITY_NONE)
|
|
|
|
clutter_actor_set_anchor_point (self, 0, 0);
|
|
|
|
else
|
2007-11-28 12:23:31 +00:00
|
|
|
{
|
2009-01-22 13:07:33 +00:00
|
|
|
clutter_anchor_coord_set_gravity (&self->priv->anchor, gravity);
|
2007-11-28 12:23:31 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
g_object_notify (G_OBJECT (self), "anchor-gravity");
|
|
|
|
g_object_notify (G_OBJECT (self), "anchor-x");
|
|
|
|
g_object_notify (G_OBJECT (self), "anchor-y");
|
|
|
|
}
|
2007-11-28 12:23:31 +00:00
|
|
|
}
|
|
|
|
|
2007-11-30 09:27:10 +00:00
|
|
|
typedef enum
|
|
|
|
{
|
|
|
|
PARSE_X,
|
|
|
|
PARSE_Y,
|
|
|
|
PARSE_WIDTH,
|
2008-03-06 04:37:00 +00:00
|
|
|
PARSE_HEIGHT,
|
|
|
|
PARSE_ANCHOR_X,
|
|
|
|
PARSE_ANCHOR_Y
|
2007-11-30 09:27:10 +00:00
|
|
|
} ParseDimension;
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
static gfloat
|
2007-11-30 14:36:07 +00:00
|
|
|
parse_units (ClutterActor *self,
|
|
|
|
ParseDimension dimension,
|
2007-11-30 09:27:10 +00:00
|
|
|
JsonNode *node)
|
|
|
|
{
|
|
|
|
GValue value = { 0, };
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat retval = 0;
|
2007-11-30 09:27:10 +00:00
|
|
|
|
|
|
|
if (JSON_NODE_TYPE (node) != JSON_NODE_VALUE)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
json_node_get_value (node, &value);
|
|
|
|
|
2009-08-12 15:02:23 +00:00
|
|
|
if (G_VALUE_HOLDS (&value, G_TYPE_INT64))
|
2007-11-30 09:27:10 +00:00
|
|
|
{
|
2009-08-12 15:02:23 +00:00
|
|
|
retval = (gfloat) g_value_get_int64 (&value);
|
2009-06-03 10:12:09 +00:00
|
|
|
}
|
2009-08-26 15:49:45 +00:00
|
|
|
else if (G_VALUE_HOLDS (&value, G_TYPE_DOUBLE))
|
2009-06-03 10:12:09 +00:00
|
|
|
{
|
2009-08-26 15:49:45 +00:00
|
|
|
retval = g_value_get_double (&value);
|
2007-11-30 09:27:10 +00:00
|
|
|
}
|
|
|
|
else if (G_VALUE_HOLDS (&value, G_TYPE_STRING))
|
|
|
|
{
|
2009-06-03 10:12:09 +00:00
|
|
|
ClutterUnits units;
|
|
|
|
gboolean res;
|
2009-05-07 17:46:36 +00:00
|
|
|
|
2009-06-03 10:12:09 +00:00
|
|
|
res = clutter_units_from_string (&units, g_value_get_string (&value));
|
|
|
|
if (res)
|
|
|
|
retval = clutter_units_to_pixels (&units);
|
|
|
|
else
|
2007-11-30 09:27:10 +00:00
|
|
|
{
|
2009-06-03 10:12:09 +00:00
|
|
|
g_warning ("Invalid value '%s': integers, strings or floating point "
|
|
|
|
"values can be used for the x, y, width and height "
|
|
|
|
"properties. Valid modifiers for strings are 'px', 'mm', "
|
|
|
|
"'pt' and 'em'.",
|
|
|
|
g_value_get_string (&value));
|
|
|
|
retval = 0;
|
2007-11-30 09:27:10 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2009-03-17 14:12:01 +00:00
|
|
|
g_warning ("Invalid value of type '%s': integers, strings of floating "
|
2008-03-06 04:37:00 +00:00
|
|
|
"point values can be used for the x, y, width, height "
|
|
|
|
"anchor-x and anchor-y properties.",
|
2007-11-30 09:27:10 +00:00
|
|
|
g_type_name (G_VALUE_TYPE (&value)));
|
|
|
|
}
|
|
|
|
|
|
|
|
g_value_unset (&value);
|
|
|
|
|
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
typedef struct {
|
|
|
|
ClutterRotateAxis axis;
|
|
|
|
|
2009-01-23 15:55:41 +00:00
|
|
|
gdouble angle;
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat center_x;
|
|
|
|
gfloat center_y;
|
|
|
|
gfloat center_z;
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
} RotationInfo;
|
|
|
|
|
2007-12-25 22:31:11 +00:00
|
|
|
static inline gboolean
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
parse_rotation_array (ClutterActor *actor,
|
2007-12-25 22:31:11 +00:00
|
|
|
JsonArray *array,
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
RotationInfo *info)
|
|
|
|
{
|
|
|
|
JsonNode *element;
|
|
|
|
|
|
|
|
if (json_array_get_length (array) != 2)
|
|
|
|
return FALSE;
|
|
|
|
|
|
|
|
/* angle */
|
|
|
|
element = json_array_get_element (array, 0);
|
|
|
|
if (JSON_NODE_TYPE (element) == JSON_NODE_VALUE)
|
2009-01-23 15:55:41 +00:00
|
|
|
info->angle = json_node_get_double (element);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
else
|
|
|
|
return FALSE;
|
|
|
|
|
|
|
|
/* center */
|
|
|
|
element = json_array_get_element (array, 1);
|
|
|
|
if (JSON_NODE_TYPE (element) == JSON_NODE_ARRAY)
|
|
|
|
{
|
2007-12-21 15:25:00 +00:00
|
|
|
JsonArray *center = json_node_get_array (element);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
2007-12-21 15:25:00 +00:00
|
|
|
if (json_array_get_length (center) != 2)
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
return FALSE;
|
|
|
|
|
|
|
|
switch (info->axis)
|
|
|
|
{
|
|
|
|
case CLUTTER_X_AXIS:
|
|
|
|
info->center_y = parse_units (actor, PARSE_Y,
|
2007-12-21 15:25:00 +00:00
|
|
|
json_array_get_element (center, 0));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
info->center_z = parse_units (actor, PARSE_Y,
|
2007-12-21 15:25:00 +00:00
|
|
|
json_array_get_element (center, 1));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
return TRUE;
|
|
|
|
|
|
|
|
case CLUTTER_Y_AXIS:
|
|
|
|
info->center_x = parse_units (actor, PARSE_X,
|
2007-12-21 15:25:00 +00:00
|
|
|
json_array_get_element (center, 0));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
info->center_z = parse_units (actor, PARSE_X,
|
2007-12-21 15:25:00 +00:00
|
|
|
json_array_get_element (center, 1));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
return TRUE;
|
|
|
|
|
|
|
|
case CLUTTER_Z_AXIS:
|
|
|
|
info->center_x = parse_units (actor, PARSE_X,
|
2007-12-21 15:25:00 +00:00
|
|
|
json_array_get_element (center, 0));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
info->center_y = parse_units (actor, PARSE_Y,
|
2007-12-21 15:25:00 +00:00
|
|
|
json_array_get_element (center, 1));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
|
|
|
static gboolean
|
|
|
|
parse_rotation (ClutterActor *actor,
|
|
|
|
JsonNode *node,
|
|
|
|
RotationInfo *info)
|
|
|
|
{
|
|
|
|
JsonArray *array;
|
|
|
|
guint len, i;
|
|
|
|
gboolean retval = FALSE;
|
|
|
|
|
|
|
|
if (JSON_NODE_TYPE (node) != JSON_NODE_ARRAY)
|
|
|
|
{
|
2009-03-17 14:12:01 +00:00
|
|
|
g_warning ("Invalid node of type '%s' found, expecting an array",
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
json_node_type_name (node));
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
|
|
|
array = json_node_get_array (node);
|
|
|
|
len = json_array_get_length (array);
|
|
|
|
|
|
|
|
for (i = 0; i < len; i++)
|
|
|
|
{
|
|
|
|
JsonNode *element = json_array_get_element (array, i);
|
|
|
|
JsonObject *object;
|
|
|
|
JsonNode *member;
|
|
|
|
|
|
|
|
if (JSON_NODE_TYPE (element) != JSON_NODE_OBJECT)
|
|
|
|
{
|
2009-03-17 14:12:01 +00:00
|
|
|
g_warning ("Invalid node of type '%s' found, expecting an object",
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
json_node_type_name (element));
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
|
|
|
object = json_node_get_object (element);
|
|
|
|
|
|
|
|
if (json_object_has_member (object, "x-axis"))
|
|
|
|
{
|
|
|
|
member = json_object_get_member (object, "x-axis");
|
|
|
|
|
|
|
|
info->axis = CLUTTER_X_AXIS;
|
|
|
|
|
|
|
|
if (JSON_NODE_TYPE (member) == JSON_NODE_VALUE)
|
|
|
|
{
|
|
|
|
info->angle = json_node_get_double (member);
|
|
|
|
retval = TRUE;
|
|
|
|
}
|
|
|
|
else if (JSON_NODE_TYPE (member) == JSON_NODE_ARRAY)
|
2007-12-25 22:31:11 +00:00
|
|
|
retval = parse_rotation_array (actor,
|
|
|
|
json_node_get_array (member),
|
|
|
|
info);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
else
|
|
|
|
retval = FALSE;
|
|
|
|
}
|
|
|
|
else if (json_object_has_member (object, "y-axis"))
|
|
|
|
{
|
|
|
|
member = json_object_get_member (object, "y-axis");
|
|
|
|
|
|
|
|
info->axis = CLUTTER_Y_AXIS;
|
|
|
|
|
|
|
|
if (JSON_NODE_TYPE (member) == JSON_NODE_VALUE)
|
|
|
|
{
|
|
|
|
info->angle = json_node_get_double (member);
|
|
|
|
retval = TRUE;
|
|
|
|
}
|
|
|
|
else if (JSON_NODE_TYPE (member) == JSON_NODE_ARRAY)
|
2007-12-25 22:31:11 +00:00
|
|
|
retval = parse_rotation_array (actor,
|
|
|
|
json_node_get_array (member),
|
|
|
|
info);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
else
|
|
|
|
retval = FALSE;
|
|
|
|
}
|
|
|
|
else if (json_object_has_member (object, "z-axis"))
|
|
|
|
{
|
|
|
|
member = json_object_get_member (object, "z-axis");
|
|
|
|
|
|
|
|
info->axis = CLUTTER_Z_AXIS;
|
|
|
|
|
|
|
|
if (JSON_NODE_TYPE (member) == JSON_NODE_VALUE)
|
|
|
|
{
|
|
|
|
info->angle = json_node_get_double (member);
|
|
|
|
retval = TRUE;
|
|
|
|
}
|
|
|
|
else if (JSON_NODE_TYPE (member) == JSON_NODE_ARRAY)
|
2007-12-25 22:31:11 +00:00
|
|
|
retval = parse_rotation_array (actor,
|
|
|
|
json_node_get_array (member),
|
|
|
|
info);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
else
|
|
|
|
retval = FALSE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
2007-11-30 09:27:10 +00:00
|
|
|
static gboolean
|
|
|
|
clutter_actor_parse_custom_node (ClutterScriptable *scriptable,
|
|
|
|
ClutterScript *script,
|
|
|
|
GValue *value,
|
|
|
|
const gchar *name,
|
|
|
|
JsonNode *node)
|
|
|
|
{
|
2007-11-30 14:36:07 +00:00
|
|
|
ClutterActor *actor = CLUTTER_ACTOR (scriptable);
|
2007-11-30 09:27:10 +00:00
|
|
|
gboolean retval = FALSE;
|
|
|
|
|
|
|
|
if ((name[0] == 'x' && name[1] == '\0') ||
|
|
|
|
(name[0] == 'y' && name[1] == '\0') ||
|
|
|
|
(strcmp (name, "width") == 0) ||
|
2008-03-06 04:37:00 +00:00
|
|
|
(strcmp (name, "height") == 0) ||
|
|
|
|
(strcmp (name, "anchor_x") == 0) ||
|
|
|
|
(strcmp (name, "anchor_y") == 0))
|
2007-11-30 09:27:10 +00:00
|
|
|
{
|
|
|
|
ParseDimension dimension;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat units;
|
2007-11-30 09:27:10 +00:00
|
|
|
|
|
|
|
if (name[0] == 'x')
|
|
|
|
dimension = PARSE_X;
|
|
|
|
else if (name[0] == 'y')
|
|
|
|
dimension = PARSE_Y;
|
|
|
|
else if (name[0] == 'w')
|
|
|
|
dimension = PARSE_WIDTH;
|
2008-03-06 04:37:00 +00:00
|
|
|
else if (name[0] == 'h')
|
2007-11-30 09:27:10 +00:00
|
|
|
dimension = PARSE_HEIGHT;
|
2008-03-06 04:37:00 +00:00
|
|
|
else if (name[0] == 'a' && name[7] == 'x')
|
|
|
|
dimension = PARSE_ANCHOR_X;
|
|
|
|
else if (name[0] == 'a' && name[7] == 'y')
|
|
|
|
dimension = PARSE_ANCHOR_Y;
|
|
|
|
else
|
|
|
|
return FALSE;
|
2007-11-30 09:27:10 +00:00
|
|
|
|
2007-11-30 14:36:07 +00:00
|
|
|
units = parse_units (actor, dimension, node);
|
2007-11-30 09:27:10 +00:00
|
|
|
|
2008-03-06 04:37:00 +00:00
|
|
|
/* convert back to pixels: all properties are pixel-based */
|
2009-08-12 15:02:23 +00:00
|
|
|
g_value_init (value, G_TYPE_FLOAT);
|
|
|
|
g_value_set_float (value, units);
|
2007-11-30 09:27:10 +00:00
|
|
|
|
|
|
|
retval = TRUE;
|
|
|
|
}
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
else if (strcmp (name, "rotation") == 0)
|
|
|
|
{
|
|
|
|
RotationInfo *info;
|
|
|
|
|
|
|
|
info = g_slice_new0 (RotationInfo);
|
|
|
|
retval = parse_rotation (actor, node, info);
|
|
|
|
|
|
|
|
if (retval)
|
|
|
|
{
|
|
|
|
g_value_init (value, G_TYPE_POINTER);
|
|
|
|
g_value_set_pointer (value, info);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
g_slice_free (RotationInfo, info);
|
|
|
|
}
|
2007-11-30 09:27:10 +00:00
|
|
|
|
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
static void
|
|
|
|
clutter_actor_set_custom_property (ClutterScriptable *scriptable,
|
|
|
|
ClutterScript *script,
|
|
|
|
const gchar *name,
|
|
|
|
const GValue *value)
|
|
|
|
{
|
2009-06-03 10:12:09 +00:00
|
|
|
#ifdef CLUTTER_ENABLE_DEBUG
|
|
|
|
{
|
|
|
|
gchar *tmp = g_strdup_value_contents (value);
|
|
|
|
|
|
|
|
CLUTTER_NOTE (SCRIPT,
|
|
|
|
"in ClutterActor::set_custom_property('%s') = %s",
|
|
|
|
name,
|
|
|
|
tmp);
|
|
|
|
|
|
|
|
g_free (tmp);
|
|
|
|
}
|
|
|
|
#endif
|
2008-04-25 13:37:36 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
if (strcmp (name, "rotation") == 0)
|
|
|
|
{
|
|
|
|
RotationInfo *info;
|
|
|
|
|
|
|
|
if (!G_VALUE_HOLDS (value, G_TYPE_POINTER))
|
|
|
|
return;
|
|
|
|
|
|
|
|
info = g_value_get_pointer (value);
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_set_rotation (CLUTTER_ACTOR (scriptable),
|
|
|
|
info->axis, info->angle,
|
|
|
|
info->center_x,
|
|
|
|
info->center_y,
|
|
|
|
info->center_z);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
|
|
|
g_slice_free (RotationInfo, info);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
g_object_set_property (G_OBJECT (scriptable), name, value);
|
|
|
|
}
|
|
|
|
|
2007-11-30 09:27:10 +00:00
|
|
|
static void
|
|
|
|
clutter_scriptable_iface_init (ClutterScriptableIface *iface)
|
|
|
|
{
|
|
|
|
iface->parse_custom_node = clutter_actor_parse_custom_node;
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
iface->set_custom_property = clutter_actor_set_custom_property;
|
2007-11-30 09:27:10 +00:00
|
|
|
}
|
|
|
|
|
2007-12-05 14:54:15 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_transform_stage_point
|
|
|
|
* @self: A #ClutterActor
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* @x: (in): x screen coordinate of the point to unproject
|
|
|
|
* @y: (in): y screen coordinate of the point to unproject
|
|
|
|
* @x_out: (out): return location for the unprojected x coordinance
|
|
|
|
* @y_out: (out): return location for the unprojected y coordinance
|
2007-12-05 14:54:15 +00:00
|
|
|
*
|
2008-02-15 15:18:02 +00:00
|
|
|
* This function translates screen coordinates (@x, @y) to
|
|
|
|
* coordinates relative to the actor. For example, it can be used to translate
|
|
|
|
* screen events from global screen coordinates into actor-local coordinates.
|
2007-12-05 14:54:15 +00:00
|
|
|
*
|
|
|
|
* The conversion can fail, notably if the transform stack results in the
|
|
|
|
* actor being projected on the screen as a mere line.
|
|
|
|
*
|
2007-12-07 15:49:54 +00:00
|
|
|
* The conversion should not be expected to be pixel-perfect due to the
|
|
|
|
* nature of the operation. In general the error grows when the skewing
|
|
|
|
* of the actor rectangle on screen increases.
|
|
|
|
*
|
|
|
|
* Note: This function is fairly computationally intensive.
|
2007-12-05 14:54:15 +00:00
|
|
|
*
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
* Note: This function only works when the allocation is up-to-date,
|
|
|
|
* i.e. inside of paint()
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
*
|
2007-12-07 15:49:54 +00:00
|
|
|
* Return value: %TRUE if conversion was successful.
|
2007-12-05 14:54:15 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
gboolean
|
2008-12-19 12:53:57 +00:00
|
|
|
clutter_actor_transform_stage_point (ClutterActor *self,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat x,
|
|
|
|
gfloat y,
|
|
|
|
gfloat *x_out,
|
|
|
|
gfloat *y_out)
|
2007-12-05 14:54:15 +00:00
|
|
|
{
|
|
|
|
ClutterVertex v[4];
|
2009-02-23 12:55:01 +00:00
|
|
|
float ST[3][3];
|
|
|
|
float RQ[3][3];
|
2007-12-05 14:54:15 +00:00
|
|
|
int du, dv, xi, yi;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
float px, py;
|
2009-02-23 12:55:01 +00:00
|
|
|
float xf, yf, wf, det;
|
2007-12-05 14:54:15 +00:00
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), FALSE);
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
/* This implementation is based on the quad -> quad projection algorithm
|
|
|
|
* described by Paul Heckbert in:
|
|
|
|
*
|
|
|
|
* http://www.cs.cmu.edu/~ph/texfund/texfund.pdf
|
2007-12-05 14:54:15 +00:00
|
|
|
*
|
2009-02-23 12:55:01 +00:00
|
|
|
* and the sample implementation at:
|
2007-12-05 14:54:15 +00:00
|
|
|
*
|
2009-02-23 12:55:01 +00:00
|
|
|
* http://www.cs.cmu.edu/~ph/src/texfund/
|
2007-12-05 14:54:15 +00:00
|
|
|
*
|
2008-12-19 12:53:57 +00:00
|
|
|
* Our texture is a rectangle with origin [0, 0], so we are mapping from
|
|
|
|
* quad to rectangle only, which significantly simplifies things; the
|
|
|
|
* function calls have been unrolled, and most of the math is done in fixed
|
|
|
|
* point.
|
2007-12-05 14:54:15 +00:00
|
|
|
*/
|
|
|
|
|
2008-06-10 Emmanuele Bassi <ebassi@openedhand.com>
Bug #815 - Split up request, allocation, and paint box
* clutter/clutter-actor.[ch]: Rework the size allocation,
request and paint area. Now ::request_coords() is called
::allocate(), and ::query_coords() has been split into
::get_preferred_width() and ::get_preferred_height(). See
the documentation and the layout test on how to implement
a container and layout manager with the new API. (#915,
based on a patch by Havoc Pennington, Lucas Rocha and Johan
Bilien)
* clutter/clutter-clone-texture.c: Port CloneTexture to
the new size negotiation API; it just means forwarding
the requests to the parent texture.
* clutter/clutter-deprecated.h: Add deprecated and replaced
API.
* clutter/clutter-entry.c: Port Entry to the new size
negotiation API.
* clutter/clutter-group.c: Port Group to the new size
negotiation API; the semantics of the Group actor do not
change.
* clutter/clutter-label.c: Port Label to the new size
negotiation API, and vastly simplify the code.
* clutter/clutter-main.[ch]: Add API for executing a
relayout when needed.
* clutter/clutter-private.h: Add new Stage private API.
* clutter/clutter-rectangle.c: Update the get_abs_opacity()
call to get_paint_opacity().
* clutter/clutter-stage.c:
(clutter_stage_get_preferred_width),
(clutter_stage_get_preferred_height),
(clutter_stage_allocate),
(clutter_stage_class_init): Port Stage to the new size
negotiation API.
* clutter/clutter-texture.c: Port Texture to the new size
negotiation API.
* clutter/clutter-types.h: Add ClutterRequestMode enumeration.
* clutter/x11/clutter-stage-x11.c: Port the X11 stage
implementation to the new size negotiation API.
* tests/Makefile.am: Add the layout manager test case.
* tests/test-opacity.c: Update.
* tests/test-project.c: Update.
* tests/test-layout.c: Test case for a layout manager implemented
using the new size negotiation API; the layout manager handles
both transformed and untransformed children.
2008-06-10 17:07:52 +00:00
|
|
|
clutter_actor_get_abs_allocation_vertices (self, v);
|
2007-12-05 14:54:15 +00:00
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
/* Keeping these as ints simplifies the multiplication (no significant
|
|
|
|
* loss of precision here).
|
2007-12-05 14:54:15 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
du = (int) (priv->allocation.x2 - priv->allocation.x1);
|
|
|
|
dv = (int) (priv->allocation.y2 - priv->allocation.y1);
|
2007-12-05 14:54:15 +00:00
|
|
|
|
|
|
|
if (!du || !dv)
|
|
|
|
return FALSE;
|
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
#define UX2FP(x) (x)
|
2009-02-23 12:55:01 +00:00
|
|
|
#define DET2FP(a,b,c,d) (((a) * (d)) - ((b) * (c)))
|
2007-12-05 14:54:15 +00:00
|
|
|
|
2009-02-23 12:55:01 +00:00
|
|
|
/* First, find mapping from unit uv square to xy quadrilateral; this
|
2007-12-05 14:54:15 +00:00
|
|
|
* equivalent to the pmap_square_quad() functions in the sample
|
|
|
|
* implementation, which we can simplify, since our target is always
|
|
|
|
* a rectangle.
|
|
|
|
*/
|
|
|
|
px = v[0].x - v[1].x + v[3].x - v[2].x;
|
|
|
|
py = v[0].y - v[1].y + v[3].y - v[2].y;
|
|
|
|
|
|
|
|
if (!px && !py)
|
2008-12-19 12:53:57 +00:00
|
|
|
{
|
|
|
|
/* affine transform */
|
2009-02-23 12:55:01 +00:00
|
|
|
RQ[0][0] = UX2FP (v[1].x - v[0].x);
|
|
|
|
RQ[1][0] = UX2FP (v[3].x - v[1].x);
|
|
|
|
RQ[2][0] = UX2FP (v[0].x);
|
|
|
|
RQ[0][1] = UX2FP (v[1].y - v[0].y);
|
|
|
|
RQ[1][1] = UX2FP (v[3].y - v[1].y);
|
|
|
|
RQ[2][1] = UX2FP (v[0].y);
|
2007-12-05 14:54:15 +00:00
|
|
|
RQ[0][2] = 0;
|
|
|
|
RQ[1][2] = 0;
|
2009-01-20 16:20:54 +00:00
|
|
|
RQ[2][2] = 1.0;
|
2007-12-05 14:54:15 +00:00
|
|
|
}
|
|
|
|
else
|
2008-12-19 12:53:57 +00:00
|
|
|
{
|
2009-02-23 12:55:01 +00:00
|
|
|
/* projective transform */
|
2008-02-11 10:56:57 +00:00
|
|
|
double dx1, dx2, dy1, dy2, del;
|
2007-12-05 14:54:15 +00:00
|
|
|
|
2008-12-19 12:53:57 +00:00
|
|
|
dx1 = UX2FP (v[1].x - v[3].x);
|
|
|
|
dx2 = UX2FP (v[2].x - v[3].x);
|
|
|
|
dy1 = UX2FP (v[1].y - v[3].y);
|
|
|
|
dy2 = UX2FP (v[2].y - v[3].y);
|
2007-12-05 14:54:15 +00:00
|
|
|
|
2008-12-19 12:53:57 +00:00
|
|
|
del = DET2FP (dx1, dx2, dy1, dy2);
|
2007-12-05 14:54:15 +00:00
|
|
|
if (!del)
|
|
|
|
return FALSE;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The division here needs to be done in floating point for
|
|
|
|
* precisions reasons.
|
|
|
|
*/
|
2009-02-23 12:55:01 +00:00
|
|
|
RQ[0][2] = (DET2FP (UX2FP (px), dx2, UX2FP (py), dy2) / del);
|
|
|
|
RQ[1][2] = (DET2FP (dx1, UX2FP (px), dy1, UX2FP (py)) / del);
|
|
|
|
RQ[1][2] = (DET2FP (dx1, UX2FP (px), dy1, UX2FP (py)) / del);
|
2009-01-20 16:20:54 +00:00
|
|
|
RQ[2][2] = 1.0;
|
2009-02-23 12:55:01 +00:00
|
|
|
RQ[0][0] = UX2FP (v[1].x - v[0].x) + (RQ[0][2] * UX2FP (v[1].x));
|
|
|
|
RQ[1][0] = UX2FP (v[2].x - v[0].x) + (RQ[1][2] * UX2FP (v[2].x));
|
|
|
|
RQ[2][0] = UX2FP (v[0].x);
|
|
|
|
RQ[0][1] = UX2FP (v[1].y - v[0].y) + (RQ[0][2] * UX2FP (v[1].y));
|
|
|
|
RQ[1][1] = UX2FP (v[2].y - v[0].y) + (RQ[1][2] * UX2FP (v[2].y));
|
|
|
|
RQ[2][1] = UX2FP (v[0].y);
|
2007-12-05 14:54:15 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Now combine with transform from our rectangle (u0,v0,u1,v1) to unit
|
|
|
|
* square. Since our rectangle is based at 0,0 we only need to scale.
|
|
|
|
*/
|
|
|
|
RQ[0][0] /= du;
|
|
|
|
RQ[1][0] /= dv;
|
|
|
|
RQ[0][1] /= du;
|
|
|
|
RQ[1][1] /= dv;
|
|
|
|
RQ[0][2] /= du;
|
|
|
|
RQ[1][2] /= dv;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Now RQ is transform from uv rectangle to xy quadrilateral; we need an
|
|
|
|
* inverse of that.
|
|
|
|
*/
|
2009-02-23 12:55:01 +00:00
|
|
|
ST[0][0] = DET2FP (RQ[1][1], RQ[1][2], RQ[2][1], RQ[2][2]);
|
|
|
|
ST[1][0] = DET2FP (RQ[1][2], RQ[1][0], RQ[2][2], RQ[2][0]);
|
|
|
|
ST[2][0] = DET2FP (RQ[1][0], RQ[1][1], RQ[2][0], RQ[2][1]);
|
|
|
|
ST[0][1] = DET2FP (RQ[2][1], RQ[2][2], RQ[0][1], RQ[0][2]);
|
|
|
|
ST[1][1] = DET2FP (RQ[2][2], RQ[2][0], RQ[0][2], RQ[0][0]);
|
|
|
|
ST[2][1] = DET2FP (RQ[2][0], RQ[2][1], RQ[0][0], RQ[0][1]);
|
|
|
|
ST[0][2] = DET2FP (RQ[0][1], RQ[0][2], RQ[1][1], RQ[1][2]);
|
|
|
|
ST[1][2] = DET2FP (RQ[0][2], RQ[0][0], RQ[1][2], RQ[1][0]);
|
|
|
|
ST[2][2] = DET2FP (RQ[0][0], RQ[0][1], RQ[1][0], RQ[1][1]);
|
2007-12-05 14:54:15 +00:00
|
|
|
|
|
|
|
/*
|
2009-02-23 12:55:01 +00:00
|
|
|
* Check the resulting matrix is OK.
|
2007-12-05 14:54:15 +00:00
|
|
|
*/
|
2009-02-23 12:55:01 +00:00
|
|
|
det = (RQ[0][0] * ST[0][0])
|
|
|
|
+ (RQ[0][1] * ST[0][1])
|
|
|
|
+ (RQ[0][2] * ST[0][2]);
|
2007-12-05 14:54:15 +00:00
|
|
|
if (!det)
|
|
|
|
return FALSE;
|
|
|
|
|
|
|
|
/*
|
2008-12-19 12:53:57 +00:00
|
|
|
* Now transform our point with the ST matrix; the notional w
|
|
|
|
* coordinate is 1, hence the last part is simply added.
|
2007-12-05 14:54:15 +00:00
|
|
|
*/
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
xi = (int) x;
|
|
|
|
yi = (int) y;
|
2007-12-05 14:54:15 +00:00
|
|
|
|
2008-12-19 12:53:57 +00:00
|
|
|
xf = xi * ST[0][0] + yi * ST[1][0] + ST[2][0];
|
|
|
|
yf = xi * ST[0][1] + yi * ST[1][1] + ST[2][1];
|
|
|
|
wf = xi * ST[0][2] + yi * ST[1][2] + ST[2][2];
|
2007-12-05 14:54:15 +00:00
|
|
|
|
2008-02-04 15:11:29 +00:00
|
|
|
if (x_out)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*x_out = xf / wf;
|
2008-12-19 12:53:57 +00:00
|
|
|
|
2008-02-04 15:11:29 +00:00
|
|
|
if (y_out)
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
*y_out = yf / wf;
|
2007-12-05 14:54:15 +00:00
|
|
|
|
2008-12-19 12:53:57 +00:00
|
|
|
#undef UX2FP
|
2009-02-23 12:55:01 +00:00
|
|
|
#undef DET2FP
|
2007-12-05 14:54:15 +00:00
|
|
|
|
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/*
|
2007-10-25 14:36:37 +00:00
|
|
|
* ClutterGeometry
|
2006-06-13 13:17:45 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
static ClutterGeometry*
|
|
|
|
clutter_geometry_copy (const ClutterGeometry *geometry)
|
|
|
|
{
|
2008-04-28 15:32:42 +00:00
|
|
|
return g_slice_dup (ClutterGeometry, geometry);
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
2007-10-25 14:36:37 +00:00
|
|
|
static void
|
|
|
|
clutter_geometry_free (ClutterGeometry *geometry)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (G_LIKELY (geometry != NULL))
|
2007-10-25 14:36:37 +00:00
|
|
|
g_slice_free (ClutterGeometry, geometry);
|
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
GType
|
|
|
|
clutter_geometry_get_type (void)
|
|
|
|
{
|
|
|
|
static GType our_type = 0;
|
|
|
|
|
2007-11-28 14:16:37 +00:00
|
|
|
if (G_UNLIKELY (our_type == 0))
|
|
|
|
our_type =
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
g_boxed_type_register_static (I_("ClutterGeometry"),
|
2007-11-28 14:16:37 +00:00
|
|
|
(GBoxedCopyFunc) clutter_geometry_copy,
|
|
|
|
(GBoxedFreeFunc) clutter_geometry_free);
|
2006-06-13 13:17:45 +00:00
|
|
|
|
|
|
|
return our_type;
|
|
|
|
}
|
|
|
|
|
2007-05-31 09:29:12 +00:00
|
|
|
/*
|
|
|
|
* ClutterVertices
|
|
|
|
*/
|
|
|
|
|
2009-06-16 11:47:19 +00:00
|
|
|
/**
|
|
|
|
* clutter_vertex_new:
|
|
|
|
* @x: X coordinate
|
|
|
|
* @y: Y coordinate
|
|
|
|
* @z: Z coordinate
|
|
|
|
*
|
|
|
|
* Creates a new #ClutterVertex for the point in 3D space
|
|
|
|
* identified by the 3 coordinates @x, @y, @z
|
|
|
|
*
|
|
|
|
* Return value: the newly allocate #ClutterVertex. Use
|
|
|
|
* clutter_vertex_free() to free the resources
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
ClutterVertex *
|
|
|
|
clutter_vertex_new (gfloat x,
|
|
|
|
gfloat y,
|
|
|
|
gfloat z)
|
|
|
|
{
|
|
|
|
ClutterVertex *vertex;
|
|
|
|
|
|
|
|
vertex = g_slice_new (ClutterVertex);
|
|
|
|
vertex->x = x;
|
|
|
|
vertex->y = y;
|
|
|
|
vertex->z = z;
|
|
|
|
|
|
|
|
return vertex;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_vertex_copy:
|
|
|
|
* @vertex: a #ClutterVertex
|
|
|
|
*
|
|
|
|
* Copies @vertex
|
|
|
|
*
|
|
|
|
* Return value: a newly allocated copy of #ClutterVertex. Use
|
|
|
|
* clutter_vertex_free() to free the allocated resources
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
ClutterVertex *
|
2007-06-12 11:42:29 +00:00
|
|
|
clutter_vertex_copy (const ClutterVertex *vertex)
|
2007-05-31 09:29:12 +00:00
|
|
|
{
|
2009-06-16 11:47:19 +00:00
|
|
|
if (G_LIKELY (vertex != NULL))
|
|
|
|
return g_slice_dup (ClutterVertex, vertex);
|
|
|
|
|
|
|
|
return NULL;
|
2007-05-31 09:29:12 +00:00
|
|
|
}
|
|
|
|
|
2009-06-16 11:47:19 +00:00
|
|
|
/**
|
2009-06-23 14:23:53 +00:00
|
|
|
* clutter_vertex_free:
|
2009-06-16 11:47:19 +00:00
|
|
|
* @vertex: a #ClutterVertex
|
|
|
|
*
|
|
|
|
* Frees a #ClutterVertex allocated using clutter_vertex_copy()
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
2007-10-25 14:36:37 +00:00
|
|
|
clutter_vertex_free (ClutterVertex *vertex)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (G_UNLIKELY (vertex != NULL))
|
2007-10-25 14:36:37 +00:00
|
|
|
g_slice_free (ClutterVertex, vertex);
|
|
|
|
}
|
|
|
|
|
2009-06-16 11:47:19 +00:00
|
|
|
/**
|
|
|
|
* clutter_vertex_equal:
|
|
|
|
* @vertex_a: a #ClutterVertex
|
|
|
|
* @vertex_b: a #ClutterVertex
|
|
|
|
*
|
|
|
|
* Compares @vertex_a and @vertex_b for equality
|
|
|
|
*
|
|
|
|
* Return value: %TRUE if the passed #ClutterVertex are equal
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_vertex_equal (const ClutterVertex *vertex_a,
|
|
|
|
const ClutterVertex *vertex_b)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (vertex_a != NULL && vertex_b != NULL, FALSE);
|
|
|
|
|
|
|
|
if (vertex_a == vertex_b)
|
|
|
|
return TRUE;
|
|
|
|
|
|
|
|
return vertex_a->x == vertex_b->x &&
|
|
|
|
vertex_a->y == vertex_b->y &&
|
|
|
|
vertex_a->z == vertex_b->z;
|
|
|
|
}
|
|
|
|
|
2007-05-31 09:29:12 +00:00
|
|
|
GType
|
2007-06-12 11:42:29 +00:00
|
|
|
clutter_vertex_get_type (void)
|
2007-05-31 09:29:12 +00:00
|
|
|
{
|
|
|
|
static GType our_type = 0;
|
|
|
|
|
2007-11-28 14:16:37 +00:00
|
|
|
if (G_UNLIKELY (our_type == 0))
|
|
|
|
our_type =
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
g_boxed_type_register_static (I_("ClutterVertex"),
|
2007-11-28 14:16:37 +00:00
|
|
|
(GBoxedCopyFunc) clutter_vertex_copy,
|
|
|
|
(GBoxedFreeFunc) clutter_vertex_free);
|
2007-05-31 09:29:12 +00:00
|
|
|
|
|
|
|
return our_type;
|
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
/*
|
|
|
|
* ClutterActorBox
|
|
|
|
*/
|
2009-06-16 11:47:19 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_box_new:
|
|
|
|
* @x_1: X coordinate of the top left point
|
|
|
|
* @y_1: Y coordinate of the top left point
|
|
|
|
* @x_2: X coordinate of the bottom right point
|
|
|
|
* @y_2: Y coordinate of the bottom right point
|
|
|
|
*
|
|
|
|
* Allocates a new #ClutterActorBox using the passed coordinates
|
|
|
|
* for the top left and bottom right points
|
|
|
|
*
|
|
|
|
* Return value: the newly allocated #ClutterActorBox. Use
|
|
|
|
* clutter_actor_box_free() to free the resources
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
ClutterActorBox *
|
|
|
|
clutter_actor_box_new (gfloat x_1,
|
|
|
|
gfloat y_1,
|
|
|
|
gfloat x_2,
|
|
|
|
gfloat y_2)
|
|
|
|
{
|
|
|
|
ClutterActorBox *box;
|
|
|
|
|
|
|
|
box = g_slice_new (ClutterActorBox);
|
|
|
|
box->x1 = x_1;
|
|
|
|
box->y1 = y_1;
|
|
|
|
box->x2 = x_2;
|
|
|
|
box->y2 = y_2;
|
|
|
|
|
|
|
|
return box;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_box_copy:
|
|
|
|
* @box: a #ClutterActorBox
|
|
|
|
*
|
|
|
|
* Copies @box
|
|
|
|
*
|
|
|
|
* Return value: a newly allocated copy of #ClutterActorBox. Use
|
|
|
|
* clutter_actor_box_free() to free the allocated resources
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
ClutterActorBox *
|
2006-06-13 13:17:45 +00:00
|
|
|
clutter_actor_box_copy (const ClutterActorBox *box)
|
|
|
|
{
|
2009-06-16 11:47:19 +00:00
|
|
|
if (G_LIKELY (box != NULL))
|
|
|
|
return g_slice_dup (ClutterActorBox, box);
|
|
|
|
|
|
|
|
return NULL;
|
2006-06-13 13:17:45 +00:00
|
|
|
}
|
|
|
|
|
2009-06-16 11:47:19 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_box_free:
|
|
|
|
* @box: a #ClutterActorBox
|
|
|
|
*
|
|
|
|
* Frees a #ClutterActorBox allocated using clutter_actor_box_new()
|
|
|
|
* or clutter_actor_box_copy()
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
2007-10-25 14:36:37 +00:00
|
|
|
clutter_actor_box_free (ClutterActorBox *box)
|
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (G_LIKELY (box != NULL))
|
2007-10-25 14:36:37 +00:00
|
|
|
g_slice_free (ClutterActorBox, box);
|
|
|
|
}
|
|
|
|
|
2006-06-13 13:17:45 +00:00
|
|
|
GType
|
|
|
|
clutter_actor_box_get_type (void)
|
|
|
|
{
|
|
|
|
static GType our_type = 0;
|
|
|
|
|
2007-11-28 14:16:37 +00:00
|
|
|
if (G_UNLIKELY (our_type == 0))
|
|
|
|
our_type =
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
g_boxed_type_register_static (I_("ClutterActorBox"),
|
2007-11-28 14:16:37 +00:00
|
|
|
(GBoxedCopyFunc) clutter_actor_box_copy,
|
|
|
|
(GBoxedFreeFunc) clutter_actor_box_free);
|
2006-06-13 13:17:45 +00:00
|
|
|
return our_type;
|
|
|
|
}
|
2007-12-03 16:29:18 +00:00
|
|
|
|
2009-06-16 11:47:19 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_box_equal:
|
|
|
|
* @box_a: a #ClutterActorBox
|
|
|
|
* @box_b: a #ClutterActorBox
|
|
|
|
*
|
|
|
|
* Checks @box_a and @box_b for equality
|
|
|
|
*
|
|
|
|
* Return value: %TRUE if the passed #ClutterActorBox are equal
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_box_equal (const ClutterActorBox *box_a,
|
|
|
|
const ClutterActorBox *box_b)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (box_a != NULL && box_b != NULL, FALSE);
|
|
|
|
|
|
|
|
if (box_a == box_b)
|
|
|
|
return TRUE;
|
|
|
|
|
|
|
|
return box_a->x1 == box_b->x1 && box_a->y1 == box_b->y1 &&
|
|
|
|
box_a->x2 == box_b->x2 && box_a->y2 == box_b->y2;
|
|
|
|
}
|
|
|
|
|
2009-06-16 15:30:41 +00:00
|
|
|
/**
|
2009-06-23 14:23:53 +00:00
|
|
|
* clutter_actor_box_get_x:
|
2009-06-16 15:30:41 +00:00
|
|
|
* @box: a #ClutterActorBox
|
|
|
|
*
|
|
|
|
* Retrieves the X coordinate of the origin of @box
|
|
|
|
*
|
|
|
|
* Return value: the X coordinate of the origin
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
gfloat
|
|
|
|
clutter_actor_box_get_x (const ClutterActorBox *box)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (box != NULL, 0.);
|
|
|
|
|
|
|
|
return box->x1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2009-06-23 14:23:53 +00:00
|
|
|
* clutter_actor_box_get_y:
|
2009-06-16 15:30:41 +00:00
|
|
|
* @box: a #ClutterActorBox
|
|
|
|
*
|
|
|
|
* Retrieves the Y coordinate of the origin of @box
|
|
|
|
*
|
|
|
|
* Return value: the Y coordinate of the origin
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
gfloat
|
|
|
|
clutter_actor_box_get_y (const ClutterActorBox *box)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (box != NULL, 0.);
|
|
|
|
|
|
|
|
return box->y1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2009-06-23 14:23:53 +00:00
|
|
|
* clutter_actor_box_get_width:
|
2009-06-16 15:30:41 +00:00
|
|
|
* @box: a #ClutterActorBox
|
|
|
|
*
|
|
|
|
* Retrieves the width of the @box
|
|
|
|
*
|
|
|
|
* Return value: the width of the box
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
gfloat
|
|
|
|
clutter_actor_box_get_width (const ClutterActorBox *box)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (box != NULL, 0.);
|
|
|
|
|
|
|
|
return box->x2 - box->x1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2009-06-23 14:23:53 +00:00
|
|
|
* clutter_actor_box_get_height:
|
2009-06-16 15:30:41 +00:00
|
|
|
* @box: a #ClutterActorBox
|
|
|
|
*
|
|
|
|
* Retrieves the height of the @box
|
|
|
|
*
|
|
|
|
* Return value: the height of the box
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
gfloat
|
|
|
|
clutter_actor_box_get_height (const ClutterActorBox *box)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (box != NULL, 0.);
|
|
|
|
|
|
|
|
return box->y2 - box->y1;
|
|
|
|
}
|
|
|
|
|
2009-06-16 11:47:19 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_box_get_origin:
|
|
|
|
* @box: a #ClutterActorBox
|
2009-07-02 13:21:16 +00:00
|
|
|
* @x: (out) (allow-none): return location for the X coordinate, or %NULL
|
|
|
|
* @y: (out) (allow-none): return location for the Y coordinate, or %NULL
|
2009-06-16 11:47:19 +00:00
|
|
|
*
|
|
|
|
* Retrieves the origin of @box
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_box_get_origin (const ClutterActorBox *box,
|
|
|
|
gfloat *x,
|
|
|
|
gfloat *y)
|
|
|
|
{
|
|
|
|
g_return_if_fail (box != NULL);
|
|
|
|
|
|
|
|
if (x)
|
|
|
|
*x = box->x1;
|
|
|
|
|
|
|
|
if (y)
|
|
|
|
*y = box->y1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_box_get_size:
|
|
|
|
* @box: a #ClutterActorBox
|
2009-07-02 13:21:16 +00:00
|
|
|
* @width: (out) (allow-none): return location for the width, or %NULL
|
|
|
|
* @height: (out) (allow-none): return location for the height, or %NULL
|
2009-06-16 11:47:19 +00:00
|
|
|
*
|
|
|
|
* Retrieves the size of @box
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_box_get_size (const ClutterActorBox *box,
|
|
|
|
gfloat *width,
|
|
|
|
gfloat *height)
|
|
|
|
{
|
|
|
|
g_return_if_fail (box != NULL);
|
|
|
|
|
|
|
|
if (width)
|
|
|
|
*width = box->x2 - box->x1;
|
|
|
|
|
|
|
|
if (height)
|
|
|
|
*height = box->y2 - box->y1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_box_get_area:
|
|
|
|
* @box: a #ClutterActorBox
|
|
|
|
*
|
|
|
|
* Retrieves the area of @box
|
|
|
|
*
|
|
|
|
* Return value: the area of a #ClutterActorBox, in pixels
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
gfloat
|
|
|
|
clutter_actor_box_get_area (const ClutterActorBox *box)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (box != NULL, 0.);
|
|
|
|
|
|
|
|
return (box->x2 - box->x1) * (box->y2 - box->y1);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_box_contains:
|
|
|
|
* @box: a #ClutterActorBox
|
|
|
|
* @x: X coordinate of the point
|
|
|
|
* @y: Y coordinate of the point
|
|
|
|
*
|
|
|
|
* Checks whether a point with @x, @y coordinates is contained
|
|
|
|
* withing @box
|
|
|
|
*
|
|
|
|
* Return value: %TRUE if the point is contained by the #ClutterActorBox
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_box_contains (const ClutterActorBox *box,
|
|
|
|
gfloat x,
|
|
|
|
gfloat y)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (box != NULL, FALSE);
|
|
|
|
|
|
|
|
return (x > box->x1 && x < box->x2) &&
|
|
|
|
(y > box->y1 && y < box->y2);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_box_from_vertices:
|
|
|
|
* @box: a #ClutterActorBox
|
|
|
|
* @verts: (array fixed-size=4): array of four #ClutterVertex
|
|
|
|
*
|
|
|
|
* Calculates the bounding box represented by the four vertices; for details
|
|
|
|
* of the vertex array see clutter_actor_get_abs_allocation_vertices().
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_box_from_vertices (ClutterActorBox *box,
|
|
|
|
const ClutterVertex verts[])
|
|
|
|
{
|
|
|
|
gfloat x_1, x_2, y_1, y_2;
|
|
|
|
|
|
|
|
g_return_if_fail (box != NULL);
|
|
|
|
g_return_if_fail (verts != NULL);
|
|
|
|
|
|
|
|
/* 4-way min/max */
|
|
|
|
x_1 = verts[0].x;
|
|
|
|
y_1 = verts[0].y;
|
|
|
|
|
|
|
|
if (verts[1].x < x_1)
|
|
|
|
x_1 = verts[1].x;
|
|
|
|
|
|
|
|
if (verts[2].x < x_1)
|
|
|
|
x_1 = verts[2].x;
|
|
|
|
|
|
|
|
if (verts[3].x < x_1)
|
|
|
|
x_1 = verts[3].x;
|
|
|
|
|
|
|
|
if (verts[1].y < y_1)
|
|
|
|
y_1 = verts[1].y;
|
|
|
|
|
|
|
|
if (verts[2].y < y_1)
|
|
|
|
y_1 = verts[2].y;
|
|
|
|
|
|
|
|
if (verts[3].y < y_1)
|
|
|
|
y_1 = verts[3].y;
|
|
|
|
|
|
|
|
x_2 = verts[0].x;
|
|
|
|
y_2 = verts[0].y;
|
|
|
|
|
|
|
|
if (verts[1].x > x_2)
|
|
|
|
x_2 = verts[1].x;
|
|
|
|
|
|
|
|
if (verts[2].x > x_2)
|
|
|
|
x_2 = verts[2].x;
|
|
|
|
|
|
|
|
if (verts[3].x > x_2)
|
|
|
|
x_2 = verts[3].x;
|
|
|
|
|
|
|
|
if (verts[1].y > y_2)
|
|
|
|
y_2 = verts[1].y;
|
|
|
|
|
|
|
|
if (verts[2].y > y_2)
|
|
|
|
y_2 = verts[2].y;
|
|
|
|
|
|
|
|
if (verts[3].y > y_2)
|
|
|
|
y_2 = verts[3].y;
|
|
|
|
|
|
|
|
box->x1 = x_1;
|
|
|
|
box->x2 = x_2;
|
|
|
|
box->y1 = y_1;
|
|
|
|
box->y2 = y_2;
|
|
|
|
}
|
|
|
|
|
2009-12-11 23:48:58 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_box_interpolate:
|
|
|
|
* @initial: the initial #ClutterActorBox
|
|
|
|
* @final: the final #ClutterActorBox
|
|
|
|
* @progress: the interpolation progress
|
|
|
|
* @result: (out): return location for the interpolation
|
|
|
|
*
|
|
|
|
* Interpolates between @initial and @final #ClutterActorBox<!-- -->es
|
|
|
|
* using @progress
|
|
|
|
*
|
|
|
|
* Since: 1.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_box_interpolate (const ClutterActorBox *initial,
|
|
|
|
const ClutterActorBox *final,
|
|
|
|
gdouble progress,
|
|
|
|
ClutterActorBox *result)
|
|
|
|
{
|
|
|
|
g_return_if_fail (initial != NULL);
|
|
|
|
g_return_if_fail (final != NULL);
|
|
|
|
g_return_if_fail (result != NULL);
|
|
|
|
|
|
|
|
result->x1 = initial->x1 + (final->x1 - initial->x1) * progress;
|
|
|
|
result->y1 = initial->y1 + (final->y1 - initial->y1) * progress;
|
|
|
|
result->x2 = initial->x2 + (final->x2 - initial->x2) * progress;
|
|
|
|
result->y2 = initial->y2 + (final->y2 - initial->y2) * progress;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_box_clamp_to_pixel:
|
|
|
|
* @box: (inout): the #ClutterActorBox to clamp
|
|
|
|
*
|
|
|
|
* Clamps the components of @box to the nearest integer
|
|
|
|
*
|
|
|
|
* Since: 1.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_box_clamp_to_pixel (ClutterActorBox *box)
|
|
|
|
{
|
|
|
|
g_return_if_fail (box != NULL);
|
|
|
|
|
|
|
|
box->x1 = floorf (box->x1 + 0.5);
|
|
|
|
box->y1 = floorf (box->y1 + 0.5);
|
|
|
|
box->x2 = floorf (box->x2 + 0.5);
|
|
|
|
box->y2 = floorf (box->y2 + 0.5);
|
|
|
|
}
|
|
|
|
|
2007-12-03 16:29:18 +00:00
|
|
|
/******************************************************************************/
|
|
|
|
|
|
|
|
struct _ShaderData
|
|
|
|
{
|
|
|
|
ClutterShader *shader;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
|
|
|
/* list of values that should be set on the shader
|
|
|
|
* before each paint cycle
|
|
|
|
*/
|
|
|
|
GHashTable *value_hash;
|
2007-12-03 16:29:18 +00:00
|
|
|
};
|
|
|
|
|
2008-11-18 15:08:40 +00:00
|
|
|
static void
|
|
|
|
shader_value_free (gpointer data)
|
|
|
|
{
|
|
|
|
GValue *var = data;
|
|
|
|
g_value_unset (var);
|
|
|
|
g_slice_free (GValue, var);
|
|
|
|
}
|
|
|
|
|
2007-12-03 16:29:18 +00:00
|
|
|
static void
|
|
|
|
destroy_shader_data (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *actor_priv = self->priv;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
ShaderData *shader_data = actor_priv->shader_data;
|
2007-12-03 16:29:18 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (shader_data == NULL)
|
2007-12-03 16:29:18 +00:00
|
|
|
return;
|
|
|
|
|
|
|
|
if (shader_data->shader)
|
|
|
|
{
|
|
|
|
g_object_unref (shader_data->shader);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
shader_data->shader = NULL;
|
2007-12-03 16:29:18 +00:00
|
|
|
}
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
2008-11-18 15:08:40 +00:00
|
|
|
if (shader_data->value_hash)
|
2007-12-03 16:29:18 +00:00
|
|
|
{
|
2008-11-18 15:08:40 +00:00
|
|
|
g_hash_table_destroy (shader_data->value_hash);
|
|
|
|
shader_data->value_hash = NULL;
|
2007-12-03 16:29:18 +00:00
|
|
|
}
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
2009-05-20 16:26:39 +00:00
|
|
|
g_slice_free (ShaderData, shader_data);
|
2007-12-03 16:29:18 +00:00
|
|
|
actor_priv->shader_data = NULL;
|
|
|
|
}
|
|
|
|
|
2008-02-04 11:45:12 +00:00
|
|
|
|
2007-12-15 14:36:27 +00:00
|
|
|
/**
|
2008-02-04 11:45:12 +00:00
|
|
|
* clutter_actor_get_shader:
|
2007-12-15 14:36:27 +00:00
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
2008-02-04 11:45:12 +00:00
|
|
|
* Queries the currently set #ClutterShader on @self.
|
|
|
|
*
|
2009-07-02 13:21:16 +00:00
|
|
|
* Return value: (transfer none): The currently set #ClutterShader
|
|
|
|
* or %NULL if no shader is set.
|
2008-02-04 11:45:12 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
ClutterShader *
|
|
|
|
clutter_actor_get_shader (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *actor_priv;
|
|
|
|
ShaderData *shader_data;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), FALSE);
|
|
|
|
|
|
|
|
actor_priv = self->priv;
|
|
|
|
shader_data = actor_priv->shader_data;
|
|
|
|
|
2009-05-20 16:26:39 +00:00
|
|
|
if (shader_data == NULL)
|
2008-02-04 22:35:38 +00:00
|
|
|
return NULL;
|
|
|
|
|
2008-02-04 11:45:12 +00:00
|
|
|
return shader_data->shader;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_set_shader:
|
|
|
|
* @self: a #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @shader: (allow-none): a #ClutterShader or %NULL to unset the shader.
|
2008-02-04 11:45:12 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* Sets the #ClutterShader to be used when rendering @self.
|
2009-05-20 16:26:39 +00:00
|
|
|
*
|
2008-02-15 14:39:25 +00:00
|
|
|
* If @shader is %NULL it will unset any currently set shader
|
2008-02-15 15:18:02 +00:00
|
|
|
* for the actor.
|
2007-12-15 14:36:27 +00:00
|
|
|
*
|
|
|
|
* Return value: %TRUE if the shader was successfully applied
|
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
gboolean
|
2008-02-04 11:45:12 +00:00
|
|
|
clutter_actor_set_shader (ClutterActor *self,
|
|
|
|
ClutterShader *shader)
|
2007-12-03 16:29:18 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *actor_priv;
|
2008-02-04 11:45:12 +00:00
|
|
|
ShaderData *shader_data;
|
2007-12-03 16:29:18 +00:00
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), FALSE);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
g_return_val_if_fail (shader == NULL || CLUTTER_IS_SHADER (shader), FALSE);
|
2007-12-05 14:54:15 +00:00
|
|
|
|
2009-05-20 16:26:39 +00:00
|
|
|
if (shader != NULL)
|
2009-01-05 12:05:51 +00:00
|
|
|
g_object_ref (shader);
|
|
|
|
else
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
{
|
|
|
|
/* if shader passed in is NULL we destroy the shader */
|
|
|
|
destroy_shader_data (self);
|
2009-05-20 16:26:39 +00:00
|
|
|
return TRUE;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
}
|
2008-02-04 11:45:12 +00:00
|
|
|
|
2007-12-03 16:29:18 +00:00
|
|
|
actor_priv = self->priv;
|
|
|
|
shader_data = actor_priv->shader_data;
|
|
|
|
|
2009-05-20 16:26:39 +00:00
|
|
|
if (shader_data == NULL)
|
2007-12-03 16:29:18 +00:00
|
|
|
{
|
2009-05-20 16:26:39 +00:00
|
|
|
actor_priv->shader_data = shader_data = g_slice_new (ShaderData);
|
|
|
|
shader_data->shader = NULL;
|
2008-11-18 15:08:40 +00:00
|
|
|
shader_data->value_hash =
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
g_hash_table_new_full (g_str_hash, g_str_equal,
|
|
|
|
g_free,
|
2008-11-18 15:08:40 +00:00
|
|
|
shader_value_free);
|
2007-12-03 16:29:18 +00:00
|
|
|
}
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (shader_data->shader != NULL)
|
2009-01-05 12:05:51 +00:00
|
|
|
g_object_unref (shader_data->shader);
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
2009-01-05 12:05:51 +00:00
|
|
|
shader_data->shader = shader;
|
2008-02-04 11:45:12 +00:00
|
|
|
|
2009-05-20 16:26:39 +00:00
|
|
|
clutter_actor_queue_redraw (self);
|
2007-12-11 14:49:41 +00:00
|
|
|
|
2007-12-03 16:29:18 +00:00
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
|
2008-02-04 11:45:12 +00:00
|
|
|
|
2007-12-03 16:29:18 +00:00
|
|
|
static void
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
set_each_param (gpointer key,
|
|
|
|
gpointer value,
|
|
|
|
gpointer user_data)
|
2007-12-03 16:29:18 +00:00
|
|
|
{
|
2008-11-18 15:08:40 +00:00
|
|
|
ClutterShader *shader = user_data;
|
|
|
|
GValue *var = value;
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
|
2008-11-18 15:08:40 +00:00
|
|
|
clutter_shader_set_uniform (shader, (const gchar *)key, var);
|
2007-12-03 16:29:18 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_shader_pre_paint (ClutterActor *actor,
|
|
|
|
gboolean repeat)
|
|
|
|
{
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2007-12-03 16:29:18 +00:00
|
|
|
ShaderData *shader_data;
|
|
|
|
ClutterShader *shader;
|
|
|
|
ClutterMainContext *context;
|
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
priv = actor->priv;
|
|
|
|
shader_data = priv->shader_data;
|
2007-12-03 16:29:18 +00:00
|
|
|
|
|
|
|
if (!shader_data)
|
|
|
|
return;
|
|
|
|
|
2009-06-17 16:59:54 +00:00
|
|
|
context = _clutter_context_get_default ();
|
2007-12-03 16:29:18 +00:00
|
|
|
shader = shader_data->shader;
|
|
|
|
|
|
|
|
if (shader)
|
|
|
|
{
|
2007-12-11 14:49:41 +00:00
|
|
|
clutter_shader_set_is_enabled (shader, TRUE);
|
2007-12-03 16:29:18 +00:00
|
|
|
|
2008-11-18 15:08:40 +00:00
|
|
|
g_hash_table_foreach (shader_data->value_hash, set_each_param, shader);
|
2007-12-03 16:29:18 +00:00
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
if (!repeat)
|
|
|
|
context->shaders = g_slist_prepend (context->shaders, actor);
|
2007-12-03 16:29:18 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
clutter_actor_shader_post_paint (ClutterActor *actor)
|
|
|
|
{
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2007-12-03 16:29:18 +00:00
|
|
|
ShaderData *shader_data;
|
|
|
|
ClutterShader *shader;
|
|
|
|
ClutterMainContext *context;
|
|
|
|
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
priv = actor->priv;
|
|
|
|
shader_data = priv->shader_data;
|
2007-12-03 16:29:18 +00:00
|
|
|
|
|
|
|
if (!shader_data)
|
|
|
|
return;
|
|
|
|
|
2009-06-17 16:59:54 +00:00
|
|
|
context = _clutter_context_get_default ();
|
2007-12-03 16:29:18 +00:00
|
|
|
shader = shader_data->shader;
|
|
|
|
|
|
|
|
if (shader)
|
|
|
|
{
|
2007-12-11 14:49:41 +00:00
|
|
|
clutter_shader_set_is_enabled (shader, FALSE);
|
2007-12-03 16:29:18 +00:00
|
|
|
|
|
|
|
context->shaders = g_slist_remove (context->shaders, actor);
|
|
|
|
if (context->shaders)
|
|
|
|
{
|
|
|
|
/* call pre-paint again, this time with the second argument being
|
|
|
|
* TRUE, indicating that we are reapplying the shader and thus
|
|
|
|
* should not be prepended to the stack
|
|
|
|
*/
|
|
|
|
clutter_actor_shader_pre_paint (context->shaders->data, TRUE);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-12-15 14:36:27 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_shader_param:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @param: the name of the parameter
|
|
|
|
* @value: the value of the parameter
|
|
|
|
*
|
|
|
|
* Sets the value for a named parameter of the shader applied
|
|
|
|
* to @actor.
|
|
|
|
*
|
2008-11-18 15:08:40 +00:00
|
|
|
* Since: 1.0
|
2007-12-15 14:36:27 +00:00
|
|
|
*/
|
2007-12-03 16:29:18 +00:00
|
|
|
void
|
2007-12-15 14:36:27 +00:00
|
|
|
clutter_actor_set_shader_param (ClutterActor *self,
|
2007-12-03 16:29:18 +00:00
|
|
|
const gchar *param,
|
2008-11-18 15:08:40 +00:00
|
|
|
const GValue *value)
|
2007-12-03 16:29:18 +00:00
|
|
|
{
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
ClutterActorPrivate *priv;
|
2007-12-03 16:29:18 +00:00
|
|
|
ShaderData *shader_data;
|
2008-11-18 15:08:40 +00:00
|
|
|
GValue *var;
|
2007-12-03 16:29:18 +00:00
|
|
|
|
2007-12-15 14:36:27 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
g_return_if_fail (param != NULL);
|
2008-11-18 15:08:40 +00:00
|
|
|
g_return_if_fail (CLUTTER_VALUE_HOLDS_SHADER_FLOAT (value) ||
|
|
|
|
CLUTTER_VALUE_HOLDS_SHADER_INT (value) ||
|
|
|
|
CLUTTER_VALUE_HOLDS_SHADER_MATRIX (value) ||
|
|
|
|
G_VALUE_HOLDS_FLOAT (value) ||
|
|
|
|
G_VALUE_HOLDS_INT (value));
|
2007-12-05 14:54:15 +00:00
|
|
|
|
2007-12-15 14:36:27 +00:00
|
|
|
priv = self->priv;
|
2007-12-10 Emmanuele Bassi <ebassi@openedhand.com>
* clutter/clutter-actor.c:
(clutter_actor_set_property),
(clutter_actor_get_property),
(clutter_actor_class_init): Add the rotation-* properties,
controlling the rotation angle and center on each axis. (#614)
(parse_rotation_array), (parse_rotation),
(clutter_actor_parse_custom_node),
(clutter_actor_set_custom_property),
(clutter_scriptable_iface_init): Add a new "rotation" custom
property for expressing the rotation along each axis in a
compact way:
"rotation" : [
{ "x-axis" : [ <angle>, [ <y>, <z> ] ] },
{ "y-axis" : [ <angle>, [ <x>, <z> ] ] },
{ "z-axis" : [ <angle>, [ <x>, <y> ] ] }
]
(clutter_geometry_get_type),(clutter_vertex_get_type),
(clutter_actor_box_get_type): Use the I_() macro instead of
directly calling g_intern_static_string().
* clutter/clutter-entry.c (clutter_entry_request_coords):
* clutter/clutter-label.c (clutter_label_ensure_layout),
(clutter_label_request_coords): Use CLUTTER_UNITS_FROM_DEVICE()
instead of CLUTTER_UNITS_FROM_INT(), as "device" means "pixels".
* clutter/clutter-private.h: Add the I_() macro for intern
static strings.
* tests/test-script.json: Test the newly added "rotation"
custom property.
2007-12-10 11:01:10 +00:00
|
|
|
shader_data = priv->shader_data;
|
2007-12-03 16:29:18 +00:00
|
|
|
|
|
|
|
if (!shader_data)
|
|
|
|
return;
|
|
|
|
|
2008-11-18 15:08:40 +00:00
|
|
|
var = g_slice_new0 (GValue);
|
|
|
|
g_value_init (var, G_VALUE_TYPE (value));
|
|
|
|
g_value_copy (value, var);
|
|
|
|
g_hash_table_insert (shader_data->value_hash, g_strdup (param), var);
|
2008-04-15 13:55:36 +00:00
|
|
|
|
|
|
|
if (CLUTTER_ACTOR_IS_VISIBLE (self))
|
|
|
|
clutter_actor_queue_redraw (self);
|
2007-12-03 16:29:18 +00:00
|
|
|
}
|
2008-02-15 09:13:50 +00:00
|
|
|
|
2008-11-18 15:08:40 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_set_shader_param_float:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @param: the name of the parameter
|
|
|
|
* @value: the value of the parameter
|
|
|
|
*
|
|
|
|
* Sets the value for a named float parameter of the shader applied
|
|
|
|
* to @actor.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_shader_param_float (ClutterActor *self,
|
|
|
|
const gchar *param,
|
|
|
|
gfloat value)
|
|
|
|
{
|
|
|
|
GValue var = { 0, };
|
|
|
|
|
|
|
|
g_value_init (&var, G_TYPE_FLOAT);
|
|
|
|
g_value_set_float (&var, value);
|
|
|
|
|
|
|
|
clutter_actor_set_shader_param (self, param, &var);
|
|
|
|
|
|
|
|
g_value_unset (&var);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_set_shader_param_int:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @param: the name of the parameter
|
|
|
|
* @value: the value of the parameter
|
|
|
|
*
|
|
|
|
* Sets the value for a named int parameter of the shader applied to
|
|
|
|
* @actor.
|
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_shader_param_int (ClutterActor *self,
|
|
|
|
const gchar *param,
|
|
|
|
gint value)
|
|
|
|
{
|
|
|
|
GValue var = { 0, };
|
|
|
|
|
|
|
|
g_value_init (&var, G_TYPE_INT);
|
|
|
|
g_value_set_int (&var, value);
|
|
|
|
|
|
|
|
clutter_actor_set_shader_param (self, param, &var);
|
|
|
|
|
|
|
|
g_value_unset (&var);
|
|
|
|
}
|
|
|
|
|
2008-02-15 09:13:50 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_is_rotated:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
2008-02-15 09:40:41 +00:00
|
|
|
* Checks whether any rotation is applied to the actor.
|
|
|
|
*
|
|
|
|
* Return value: %TRUE if the actor is rotated.
|
2008-02-15 09:13:50 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_is_rotated (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), FALSE);
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->rxang || priv->ryang || priv->rzang)
|
|
|
|
return TRUE;
|
|
|
|
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_is_scaled:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
2008-02-15 09:40:41 +00:00
|
|
|
* Checks whether the actor is scaled in either dimension.
|
|
|
|
*
|
|
|
|
* Return value: %TRUE if the actor is scaled.
|
2008-02-15 09:13:50 +00:00
|
|
|
*
|
|
|
|
* Since: 0.6
|
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_is_scaled (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), FALSE);
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
2009-01-20 16:20:54 +00:00
|
|
|
if (priv->scale_x != 1.0 || priv->scale_y != 1.0)
|
2008-02-15 09:13:50 +00:00
|
|
|
return TRUE;
|
|
|
|
|
|
|
|
return FALSE;
|
|
|
|
}
|
|
|
|
|
2008-04-28 15:32:42 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_stage:
|
|
|
|
* @actor: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the #ClutterStage where @actor is contained.
|
|
|
|
*
|
2009-02-17 00:25:20 +00:00
|
|
|
* Return value: (transfer none): the stage containing the actor, or %NULL
|
2008-04-28 15:32:42 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
ClutterActor *
|
2008-03-28 22:50:55 +00:00
|
|
|
clutter_actor_get_stage (ClutterActor *actor)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (actor), NULL);
|
|
|
|
|
|
|
|
while (actor && !(CLUTTER_PRIVATE_FLAGS (actor) & CLUTTER_ACTOR_IS_TOPLEVEL))
|
|
|
|
actor = clutter_actor_get_parent (actor);
|
|
|
|
|
|
|
|
return actor;
|
|
|
|
}
|
2008-06-17 13:49:40 +00:00
|
|
|
|
2009-05-29 16:53:19 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_allocate_available_size:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @x: the actor's X coordinate
|
|
|
|
* @y: the actor's Y coordinate
|
|
|
|
* @available_width: the maximum available width, or -1 to use the
|
|
|
|
* actor's natural width
|
|
|
|
* @available_height: the maximum available height, or -1 to use the
|
|
|
|
* actor's natural height
|
2009-06-03 13:02:06 +00:00
|
|
|
* @flags: flags controlling the allocation
|
2009-05-29 16:53:19 +00:00
|
|
|
*
|
|
|
|
* Allocates @self taking into account the #ClutterActor<!-- -->'s
|
|
|
|
* preferred size, but limiting it to the maximum available width
|
|
|
|
* and height provided.
|
|
|
|
*
|
|
|
|
* This function will do the right thing when dealing with the
|
|
|
|
* actor's request mode.
|
|
|
|
*
|
|
|
|
* The implementation of this function is equivalent to:
|
|
|
|
*
|
|
|
|
* |[
|
|
|
|
* if (request_mode == CLUTTER_REQUEST_HEIGHT_FOR_WIDTH)
|
|
|
|
* {
|
|
|
|
* clutter_actor_get_preferred_width (self, available_height,
|
|
|
|
* &min_width,
|
|
|
|
* &natural_width);
|
|
|
|
* width = CLAMP (natural_width, min_width, available_width);
|
|
|
|
*
|
|
|
|
* clutter_actor_get_preferred_height (self, width,
|
|
|
|
* &min_height,
|
|
|
|
* &natural_height);
|
|
|
|
* height = CLAMP (natural_height, min_height, available_height);
|
|
|
|
* }
|
|
|
|
* else
|
|
|
|
* {
|
|
|
|
* clutter_actor_get_preferred_height (self, available_width,
|
|
|
|
* &min_height,
|
|
|
|
* &natural_height);
|
|
|
|
* height = CLAMP (natural_height, min_height, available_height);
|
|
|
|
*
|
|
|
|
* clutter_actor_get_preferred_width (self, height,
|
|
|
|
* &min_width,
|
|
|
|
* &natural_width);
|
|
|
|
* width = CLAMP (natural_width, min_width, available_width);
|
|
|
|
* }
|
|
|
|
*
|
|
|
|
* box.x1 = x; box.y1 = y;
|
|
|
|
* box.x2 = box.x1 + available_width;
|
|
|
|
* box.y2 = box.y1 + available_height;
|
2009-06-03 13:02:06 +00:00
|
|
|
* clutter_actor_allocate (self, &box, flags);
|
2009-05-29 16:53:19 +00:00
|
|
|
* ]|
|
|
|
|
*
|
|
|
|
* This function can be used by fluid layout managers to allocate
|
|
|
|
* an actor's preferred size without making it bigger than the area
|
|
|
|
* available for the container.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
2009-06-03 13:02:06 +00:00
|
|
|
clutter_actor_allocate_available_size (ClutterActor *self,
|
|
|
|
gfloat x,
|
|
|
|
gfloat y,
|
|
|
|
gfloat available_width,
|
|
|
|
gfloat available_height,
|
|
|
|
ClutterAllocationFlags flags)
|
2009-05-29 16:53:19 +00:00
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
gfloat width, height;
|
|
|
|
gfloat min_width, min_height;
|
|
|
|
gfloat natural_width, natural_height;
|
|
|
|
ClutterActorBox box;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
2009-06-06 13:37:41 +00:00
|
|
|
width = height = 0.0;
|
|
|
|
|
2009-05-29 16:53:19 +00:00
|
|
|
switch (priv->request_mode)
|
|
|
|
{
|
|
|
|
case CLUTTER_REQUEST_HEIGHT_FOR_WIDTH:
|
|
|
|
clutter_actor_get_preferred_width (self, available_height,
|
|
|
|
&min_width,
|
|
|
|
&natural_width);
|
|
|
|
width = CLAMP (natural_width, min_width, available_width);
|
|
|
|
|
|
|
|
clutter_actor_get_preferred_height (self, width,
|
|
|
|
&min_height,
|
|
|
|
&natural_height);
|
|
|
|
height = CLAMP (natural_height, min_height, available_height);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_REQUEST_WIDTH_FOR_HEIGHT:
|
|
|
|
clutter_actor_get_preferred_height (self, available_width,
|
|
|
|
&min_height,
|
|
|
|
&natural_height);
|
|
|
|
height = CLAMP (natural_height, min_height, available_height);
|
|
|
|
|
|
|
|
clutter_actor_get_preferred_width (self, height,
|
|
|
|
&min_width,
|
|
|
|
&natural_width);
|
|
|
|
width = CLAMP (natural_width, min_width, available_width);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
box.x1 = x;
|
|
|
|
box.y1 = y;
|
|
|
|
box.x2 = box.x1 + width;
|
|
|
|
box.y2 = box.y1 + height;
|
2009-06-03 13:02:06 +00:00
|
|
|
clutter_actor_allocate (self, &box, flags);
|
2009-05-29 16:53:19 +00:00
|
|
|
}
|
|
|
|
|
2008-06-17 13:49:40 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_allocate_preferred_size:
|
2008-06-17 15:59:08 +00:00
|
|
|
* @self: a #ClutterActor
|
2009-06-03 13:02:06 +00:00
|
|
|
* @flags: flags controlling the allocation
|
2008-06-17 13:49:40 +00:00
|
|
|
*
|
2008-06-17 15:59:08 +00:00
|
|
|
* Allocates the natural size of @self.
|
|
|
|
*
|
|
|
|
* This function is a utility call for #ClutterActor implementations
|
|
|
|
* that allocates the actor's preferred natural size. It can be used
|
2008-06-23 09:55:42 +00:00
|
|
|
* by fixed layout managers (like #ClutterGroup or so called
|
|
|
|
* 'composite actors') inside the ClutterActor::allocate
|
|
|
|
* implementation to give each child exactly how much space it
|
|
|
|
* requires.
|
2008-06-17 15:59:08 +00:00
|
|
|
*
|
|
|
|
* This function is not meant to be used by applications. It is also
|
|
|
|
* not meant to be used outside the implementation of the
|
|
|
|
* ClutterActor::allocate virtual function.
|
2008-06-17 13:49:40 +00:00
|
|
|
*
|
|
|
|
* Since: 0.8
|
|
|
|
*/
|
|
|
|
void
|
2009-06-03 13:02:06 +00:00
|
|
|
clutter_actor_allocate_preferred_size (ClutterActor *self,
|
|
|
|
ClutterAllocationFlags flags)
|
2008-06-17 13:49:40 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat actor_x, actor_y;
|
|
|
|
gfloat natural_width, natural_height;
|
2008-06-17 13:49:40 +00:00
|
|
|
ClutterActorBox actor_box;
|
|
|
|
|
2008-06-17 15:59:08 +00:00
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2009-02-18 18:54:54 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
actor_x = clutter_actor_get_x (self);
|
|
|
|
actor_y = clutter_actor_get_y (self);
|
2008-06-17 13:49:40 +00:00
|
|
|
|
2008-06-17 15:59:08 +00:00
|
|
|
clutter_actor_get_preferred_size (self,
|
2008-06-17 13:49:40 +00:00
|
|
|
NULL, NULL,
|
|
|
|
&natural_width,
|
|
|
|
&natural_height);
|
2009-02-18 18:54:54 +00:00
|
|
|
|
2008-06-17 13:49:40 +00:00
|
|
|
actor_box.x1 = actor_x;
|
|
|
|
actor_box.y1 = actor_y;
|
|
|
|
actor_box.x2 = actor_box.x1 + natural_width;
|
|
|
|
actor_box.y2 = actor_box.y1 + natural_height;
|
|
|
|
|
2009-06-03 13:02:06 +00:00
|
|
|
clutter_actor_allocate (self, &actor_box, flags);
|
2008-06-17 13:49:40 +00:00
|
|
|
}
|
2008-12-11 11:22:46 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_grab_key_focus:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Sets the key focus of the #ClutterStage including @self
|
|
|
|
* to this #ClutterActor.
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_grab_key_focus (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActor *parent;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
parent = clutter_actor_get_parent (self);
|
|
|
|
if (!parent)
|
|
|
|
return;
|
|
|
|
|
|
|
|
parent = clutter_actor_get_stage (self);
|
|
|
|
if (parent && CLUTTER_IS_STAGE (parent))
|
|
|
|
clutter_stage_set_key_focus (CLUTTER_STAGE (parent), self);
|
|
|
|
}
|
2008-12-23 14:06:55 +00:00
|
|
|
|
2008-12-23 16:59:14 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_get_pango_context:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the #PangoContext for @self. The actor's #PangoContext
|
|
|
|
* is already configured using the appropriate font map, resolution
|
|
|
|
* and font options.
|
|
|
|
*
|
2009-01-05 15:29:10 +00:00
|
|
|
* Unlike clutter_actor_create_pango_context(), this context is owend
|
|
|
|
* by the #ClutterActor and it will be updated each time the options
|
|
|
|
* stored by the #ClutterBackend change.
|
2008-12-23 16:59:14 +00:00
|
|
|
*
|
|
|
|
* You can use the returned #PangoContext to create a #PangoLayout
|
|
|
|
* and render text using cogl_pango_render_layout() to reuse the
|
|
|
|
* glyphs cache also used by Clutter.
|
|
|
|
*
|
2009-02-17 00:25:20 +00:00
|
|
|
* Return value: (transfer none): the #PangoContext for a #ClutterActor.
|
|
|
|
* The returned #PangoContext is owned by the actor and should not be
|
|
|
|
* unreferenced by the application code
|
2008-12-23 16:59:14 +00:00
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
2008-12-23 14:06:55 +00:00
|
|
|
PangoContext *
|
|
|
|
clutter_actor_get_pango_context (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
ClutterMainContext *ctx;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), NULL);
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->pango_context)
|
|
|
|
return priv->pango_context;
|
|
|
|
|
|
|
|
ctx = CLUTTER_CONTEXT ();
|
2009-01-05 15:29:10 +00:00
|
|
|
priv->pango_context = _clutter_context_get_pango_context (ctx);
|
2008-12-23 14:27:41 +00:00
|
|
|
g_object_ref (priv->pango_context);
|
2008-12-23 14:06:55 +00:00
|
|
|
|
|
|
|
return priv->pango_context;
|
|
|
|
}
|
2009-01-05 15:29:10 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_create_pango_context:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Creates a #PangoContext for the given actor. The #PangoContext
|
|
|
|
* is already configured using the appropriate font map, resolution
|
|
|
|
* and font options.
|
|
|
|
*
|
|
|
|
* See also clutter_actor_get_pango_context().
|
|
|
|
*
|
2009-03-11 13:55:01 +00:00
|
|
|
* Return value: the newly created #PangoContext. Use g_object_unref()
|
2009-01-05 15:29:10 +00:00
|
|
|
* on the returned value to deallocate its resources
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
PangoContext *
|
|
|
|
clutter_actor_create_pango_context (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterMainContext *ctx;
|
|
|
|
PangoContext *retval;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), NULL);
|
|
|
|
|
|
|
|
ctx = CLUTTER_CONTEXT ();
|
|
|
|
retval = _clutter_context_create_pango_context (ctx);
|
|
|
|
|
|
|
|
return retval;
|
|
|
|
}
|
2009-01-17 16:51:03 +00:00
|
|
|
|
2009-02-02 09:01:41 +00:00
|
|
|
/**
|
|
|
|
* clutter_actor_create_pango_layout:
|
|
|
|
* @self: a #ClutterActor
|
2009-07-02 13:21:16 +00:00
|
|
|
* @text: (allow-none) the text to set on the #PangoLayout, or %NULL
|
2009-02-02 09:01:41 +00:00
|
|
|
*
|
|
|
|
* Creates a new #PangoLayout from the same #PangoContext used
|
|
|
|
* by the #ClutterActor. The #PangoLayout is already configured
|
|
|
|
* with the font map, resolution and font options, and the
|
|
|
|
* given @text.
|
|
|
|
*
|
|
|
|
* If you want to keep around a #PangoLayout created by this
|
|
|
|
* function you will have to connect to the #ClutterBackend::font-changed
|
|
|
|
* and #ClutterBackend::resolution-changed signals, and call
|
|
|
|
* pango_layout_context_changed() in response to them.
|
|
|
|
*
|
|
|
|
* Return value: the newly created #PangoLayout. Use g_object_unref()
|
|
|
|
* when done
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
PangoLayout *
|
|
|
|
clutter_actor_create_pango_layout (ClutterActor *self,
|
|
|
|
const gchar *text)
|
|
|
|
{
|
|
|
|
PangoContext *context;
|
|
|
|
PangoLayout *layout;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), NULL);
|
|
|
|
|
|
|
|
context = clutter_actor_get_pango_context (self);
|
|
|
|
layout = pango_layout_new (context);
|
|
|
|
|
|
|
|
if (text)
|
|
|
|
pango_layout_set_text (layout, text, -1);
|
|
|
|
|
|
|
|
return layout;
|
|
|
|
}
|
|
|
|
|
2009-01-17 16:51:03 +00:00
|
|
|
/* Allows overriding the parent traversed when querying an actors paint
|
2009-01-27 16:12:30 +00:00
|
|
|
* opacity. Used by ClutterClone. */
|
2009-01-17 16:51:03 +00:00
|
|
|
void
|
|
|
|
_clutter_actor_set_opacity_parent (ClutterActor *self,
|
|
|
|
ClutterActor *parent)
|
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
self->priv->opacity_parent = parent;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Allows you to disable applying the actors model view transform during
|
2009-01-27 16:12:30 +00:00
|
|
|
* a paint. Used by ClutterClone. */
|
2009-01-17 16:51:03 +00:00
|
|
|
void
|
|
|
|
_clutter_actor_set_enable_model_view_transform (ClutterActor *self,
|
|
|
|
gboolean enable)
|
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
self->priv->enable_model_view_transform = enable;
|
|
|
|
}
|
|
|
|
|
2009-05-01 11:33:42 +00:00
|
|
|
void
|
|
|
|
_clutter_actor_set_enable_paint_unmapped (ClutterActor *self,
|
|
|
|
gboolean enable)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
priv->enable_paint_unmapped = enable;
|
|
|
|
|
|
|
|
if (priv->enable_paint_unmapped)
|
2009-06-06 15:56:08 +00:00
|
|
|
{
|
|
|
|
/* Make sure that the parents of the widget are realized first;
|
|
|
|
* otherwise checks in clutter_actor_update_map_state() will
|
|
|
|
* fail.
|
|
|
|
*/
|
|
|
|
clutter_actor_realize (self);
|
|
|
|
|
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_MAKE_MAPPED);
|
|
|
|
}
|
2009-05-01 11:33:42 +00:00
|
|
|
else
|
2009-06-06 15:56:08 +00:00
|
|
|
{
|
|
|
|
clutter_actor_update_map_state (self, MAP_STATE_MAKE_UNMAPPED);
|
|
|
|
}
|
2009-05-01 11:33:42 +00:00
|
|
|
}
|
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
static void
|
|
|
|
clutter_anchor_coord_get_units (ClutterActor *self,
|
|
|
|
const AnchorCoord *coord,
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat *x,
|
|
|
|
gfloat *y,
|
|
|
|
gfloat *z)
|
2009-01-22 13:07:33 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
if (coord->is_fractional)
|
2009-01-22 13:07:33 +00:00
|
|
|
{
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
gfloat actor_width, actor_height;
|
2009-01-22 13:07:33 +00:00
|
|
|
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_actor_get_size (self, &actor_width, &actor_height);
|
2009-01-22 13:07:33 +00:00
|
|
|
|
|
|
|
if (x)
|
|
|
|
*x = actor_width * coord->v.fraction.x;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
if (y)
|
|
|
|
*y = actor_height * coord->v.fraction.y;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
if (z)
|
|
|
|
*z = 0;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
if (x)
|
|
|
|
*x = coord->v.units.x;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
if (y)
|
|
|
|
*y = coord->v.units.y;
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
|
2009-01-22 13:07:33 +00:00
|
|
|
if (z)
|
|
|
|
*z = coord->v.units.z;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_anchor_coord_set_units (AnchorCoord *coord,
|
|
|
|
gfloat x,
|
|
|
|
gfloat y,
|
|
|
|
gfloat z)
|
2009-01-22 13:07:33 +00:00
|
|
|
{
|
|
|
|
coord->is_fractional = FALSE;
|
|
|
|
coord->v.units.x = x;
|
|
|
|
coord->v.units.y = y;
|
|
|
|
coord->v.units.z = z;
|
|
|
|
}
|
|
|
|
|
|
|
|
static ClutterGravity
|
|
|
|
clutter_anchor_coord_get_gravity (AnchorCoord *coord)
|
|
|
|
{
|
|
|
|
if (coord->is_fractional)
|
|
|
|
{
|
|
|
|
if (coord->v.fraction.x == 0.0)
|
|
|
|
{
|
|
|
|
if (coord->v.fraction.y == 0.0)
|
|
|
|
return CLUTTER_GRAVITY_NORTH_WEST;
|
|
|
|
else if (coord->v.fraction.y == 0.5)
|
|
|
|
return CLUTTER_GRAVITY_WEST;
|
|
|
|
else if (coord->v.fraction.y == 1.0)
|
|
|
|
return CLUTTER_GRAVITY_SOUTH_WEST;
|
|
|
|
else
|
|
|
|
return CLUTTER_GRAVITY_NONE;
|
|
|
|
}
|
|
|
|
else if (coord->v.fraction.x == 0.5)
|
|
|
|
{
|
|
|
|
if (coord->v.fraction.y == 0.0)
|
|
|
|
return CLUTTER_GRAVITY_NORTH;
|
|
|
|
else if (coord->v.fraction.y == 0.5)
|
|
|
|
return CLUTTER_GRAVITY_CENTER;
|
|
|
|
else if (coord->v.fraction.y == 1.0)
|
|
|
|
return CLUTTER_GRAVITY_SOUTH;
|
|
|
|
else
|
|
|
|
return CLUTTER_GRAVITY_NONE;
|
|
|
|
}
|
|
|
|
else if (coord->v.fraction.x == 1.0)
|
|
|
|
{
|
|
|
|
if (coord->v.fraction.y == 0.0)
|
|
|
|
return CLUTTER_GRAVITY_NORTH_EAST;
|
|
|
|
else if (coord->v.fraction.y == 0.5)
|
|
|
|
return CLUTTER_GRAVITY_EAST;
|
|
|
|
else if (coord->v.fraction.y == 1.0)
|
|
|
|
return CLUTTER_GRAVITY_SOUTH_EAST;
|
|
|
|
else
|
|
|
|
return CLUTTER_GRAVITY_NONE;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
return CLUTTER_GRAVITY_NONE;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
return CLUTTER_GRAVITY_NONE;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
Remove Units from the public API
With the recent change to internal floating point values, ClutterUnit
has become a redundant type, defined to be a float. All integer entry
points are being internally converted to floating point values to be
passed to the GL pipeline with the least amount of conversion.
ClutterUnit is thus exposed as just a "pixel with fractionary bits",
and not -- as users might think -- as generic, resolution and device
independent units. not that it was the case, but a definitive amount
of people was convinced it did provide this "feature", and was flummoxed
about the mere existence of this type.
So, having ClutterUnit exposed in the public API doubles the entry
points and has the following disadvantages:
- we have to maintain twice the amount of entry points in ClutterActor
- we still do an integer-to-float implicit conversion
- we introduce a weird impedance between pixels and "pixels with
fractionary bits"
- language bindings will have to choose what to bind, and resort
to manually overriding the API
+ *except* for language bindings based on GObject-Introspection, as
they cannot do manual overrides, thus will replicate the entire
set of entry points
For these reason, we should coalesces every Actor entry point for
pixels and for ClutterUnit into a single entry point taking a float,
like:
void clutter_actor_set_x (ClutterActor *self,
gfloat x);
void clutter_actor_get_size (ClutterActor *self,
gfloat *width,
gfloat *height);
gfloat clutter_actor_get_height (ClutterActor *self);
etc.
The issues I have identified are:
- we'll have a two cases of compiler warnings:
- printf() format of the return values from %d to %f
- clutter_actor_get_size() taking floats instead of unsigned ints
- we'll have a problem with varargs when passing an integer instead
of a floating point value, except on 64bit platforms where the
size of a float is the same as the size of an int
To be clear: the *intent* of the API should not change -- we still use
pixels everywhere -- but:
- we remove ambiguity in the API with regard to pixels and units
- we remove entry points we get to maintain for the whole 1.0
version of the API
- we make things simpler to bind for both manual language bindings
and automatic (gobject-introspection based) ones
- we have the simplest API possible while still exposing the
capabilities of the underlying GL implementation
2009-05-06 15:44:47 +00:00
|
|
|
clutter_anchor_coord_set_gravity (AnchorCoord *coord,
|
|
|
|
ClutterGravity gravity)
|
2009-01-22 13:07:33 +00:00
|
|
|
{
|
|
|
|
switch (gravity)
|
|
|
|
{
|
|
|
|
case CLUTTER_GRAVITY_NORTH:
|
|
|
|
coord->v.fraction.x = 0.5;
|
|
|
|
coord->v.fraction.y = 0.0;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_GRAVITY_NORTH_EAST:
|
|
|
|
coord->v.fraction.x = 1.0;
|
|
|
|
coord->v.fraction.y = 0.0;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_GRAVITY_EAST:
|
|
|
|
coord->v.fraction.x = 1.0;
|
|
|
|
coord->v.fraction.y = 0.5;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_GRAVITY_SOUTH_EAST:
|
|
|
|
coord->v.fraction.x = 1.0;
|
|
|
|
coord->v.fraction.y = 1.0;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_GRAVITY_SOUTH:
|
|
|
|
coord->v.fraction.x = 0.5;
|
|
|
|
coord->v.fraction.y = 1.0;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_GRAVITY_SOUTH_WEST:
|
|
|
|
coord->v.fraction.x = 0.0;
|
|
|
|
coord->v.fraction.y = 1.0;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_GRAVITY_WEST:
|
|
|
|
coord->v.fraction.x = 0.0;
|
|
|
|
coord->v.fraction.y = 0.5;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_GRAVITY_NORTH_WEST:
|
|
|
|
coord->v.fraction.x = 0.0;
|
|
|
|
coord->v.fraction.y = 0.0;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CLUTTER_GRAVITY_CENTER:
|
|
|
|
coord->v.fraction.x = 0.5;
|
|
|
|
coord->v.fraction.y = 0.5;
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
coord->v.fraction.x = 0.0;
|
|
|
|
coord->v.fraction.y = 0.0;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
coord->is_fractional = TRUE;
|
|
|
|
}
|
|
|
|
|
|
|
|
static gboolean
|
|
|
|
clutter_anchor_coord_is_zero (const AnchorCoord *coord)
|
|
|
|
{
|
|
|
|
if (coord->is_fractional)
|
|
|
|
return coord->v.fraction.x == 0.0 && coord->v.fraction.y == 0.0;
|
|
|
|
else
|
|
|
|
return (coord->v.units.x == 0.0
|
|
|
|
&& coord->v.units.y == 0.0
|
|
|
|
&& coord->v.units.z == 0.0);
|
|
|
|
}
|
2009-05-05 14:37:41 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_flags:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the flags set on @self
|
|
|
|
*
|
|
|
|
* Return value: a bitwise or of #ClutterActorFlags or 0
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
ClutterActorFlags
|
|
|
|
clutter_actor_get_flags (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), 0);
|
|
|
|
|
|
|
|
return self->flags;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_set_flags:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @flags: the flags to set
|
|
|
|
*
|
|
|
|
* Sets @flags on @self
|
|
|
|
*
|
|
|
|
* This function will emit notifications for the changed properties
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_flags (ClutterActor *self,
|
|
|
|
ClutterActorFlags flags)
|
|
|
|
{
|
|
|
|
ClutterActorFlags old_flags;
|
|
|
|
GObject *obj;
|
|
|
|
gboolean was_reactive_set, reactive_set;
|
|
|
|
gboolean was_realized_set, realized_set;
|
|
|
|
gboolean was_mapped_set, mapped_set;
|
|
|
|
gboolean was_visible_set, visible_set;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
if (self->flags == flags)
|
|
|
|
return;
|
|
|
|
|
|
|
|
obj = G_OBJECT (self);
|
|
|
|
g_object_freeze_notify (obj);
|
|
|
|
|
|
|
|
old_flags = self->flags;
|
|
|
|
|
|
|
|
was_reactive_set = ((old_flags & CLUTTER_ACTOR_REACTIVE) != 0);
|
|
|
|
was_realized_set = ((old_flags & CLUTTER_ACTOR_REALIZED) != 0);
|
|
|
|
was_mapped_set = ((old_flags & CLUTTER_ACTOR_MAPPED) != 0);
|
|
|
|
was_visible_set = ((old_flags & CLUTTER_ACTOR_VISIBLE) != 0);
|
|
|
|
|
|
|
|
self->flags |= flags;
|
|
|
|
|
|
|
|
reactive_set = ((self->flags & CLUTTER_ACTOR_REACTIVE) != 0);
|
|
|
|
realized_set = ((self->flags & CLUTTER_ACTOR_REALIZED) != 0);
|
|
|
|
mapped_set = ((self->flags & CLUTTER_ACTOR_MAPPED) != 0);
|
|
|
|
visible_set = ((self->flags & CLUTTER_ACTOR_VISIBLE) != 0);
|
|
|
|
|
|
|
|
if (reactive_set != was_reactive_set)
|
|
|
|
g_object_notify (obj, "reactive");
|
|
|
|
|
|
|
|
if (realized_set != was_realized_set)
|
|
|
|
g_object_notify (obj, "realized");
|
|
|
|
|
|
|
|
if (mapped_set != was_mapped_set)
|
|
|
|
g_object_notify (obj, "mapped");
|
|
|
|
|
|
|
|
if (visible_set != was_visible_set)
|
|
|
|
g_object_notify (obj, "visible");
|
|
|
|
|
|
|
|
g_object_thaw_notify (obj);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_unset_flags:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @flags: the flags to unset
|
|
|
|
*
|
|
|
|
* Unsets @flags on @self
|
|
|
|
*
|
|
|
|
* This function will emit notifications for the changed properties
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_unset_flags (ClutterActor *self,
|
|
|
|
ClutterActorFlags flags)
|
|
|
|
{
|
|
|
|
ClutterActorFlags old_flags;
|
|
|
|
GObject *obj;
|
|
|
|
gboolean was_reactive_set, reactive_set;
|
|
|
|
gboolean was_realized_set, realized_set;
|
|
|
|
gboolean was_mapped_set, mapped_set;
|
|
|
|
gboolean was_visible_set, visible_set;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
obj = G_OBJECT (self);
|
|
|
|
g_object_freeze_notify (obj);
|
|
|
|
|
|
|
|
old_flags = self->flags;
|
|
|
|
|
|
|
|
was_reactive_set = ((old_flags & CLUTTER_ACTOR_REACTIVE) != 0);
|
|
|
|
was_realized_set = ((old_flags & CLUTTER_ACTOR_REALIZED) != 0);
|
|
|
|
was_mapped_set = ((old_flags & CLUTTER_ACTOR_MAPPED) != 0);
|
|
|
|
was_visible_set = ((old_flags & CLUTTER_ACTOR_VISIBLE) != 0);
|
|
|
|
|
|
|
|
self->flags &= ~flags;
|
|
|
|
|
|
|
|
if (self->flags == old_flags)
|
|
|
|
return;
|
|
|
|
|
|
|
|
reactive_set = ((self->flags & CLUTTER_ACTOR_REACTIVE) != 0);
|
|
|
|
realized_set = ((self->flags & CLUTTER_ACTOR_REALIZED) != 0);
|
|
|
|
mapped_set = ((self->flags & CLUTTER_ACTOR_MAPPED) != 0);
|
|
|
|
visible_set = ((self->flags & CLUTTER_ACTOR_VISIBLE) != 0);
|
|
|
|
|
|
|
|
if (reactive_set != was_reactive_set)
|
|
|
|
g_object_notify (obj, "reactive");
|
|
|
|
|
|
|
|
if (realized_set != was_realized_set)
|
|
|
|
g_object_notify (obj, "realized");
|
|
|
|
|
|
|
|
if (mapped_set != was_mapped_set)
|
|
|
|
g_object_notify (obj, "mapped");
|
|
|
|
|
|
|
|
if (visible_set != was_visible_set)
|
|
|
|
g_object_notify (obj, "visible");
|
|
|
|
|
|
|
|
g_object_thaw_notify (obj);
|
|
|
|
}
|
[actor] Allow changing the transformations matrix
Currently, the transformation matrix for an actor is constructed
from scenegraph-related accessors. An actor, though, can call COGL
API to add new transformations inside the paint() implementation,
for instance:
static void
my_foo_paint (ClutterActor *a)
{
...
cogl_translate (-scroll_x, -scroll_y, 0);
...
}
Unfortunately these transformations will be completely ignored by
the scenegraph machinery; for instance, getting the actor-relative
coordinates from event coordinates is going to break badly because
of this.
In order to make the scenegraph aware of the potential of additional
transformations, we need a ::apply_transform() virtual function. This
vfunc will pass a CoglMatrix which can be used to apply additional
operations:
static void
my_foo_apply_transform (ClutterActor *a, CoglMatrix *m)
{
CLUTTER_ACTOR_CLASS (my_foo_parent_class)->apply_transform (a, m);
...
cogl_matrix_translate (m, -scroll_x, -scroll_y, 0);
...
}
The ::paint() implementation will be called with the actor already
using the newly applied transformation matrix, as expected:
static void
my_foo_paint (ClutterActor *a)
{
...
}
The ::apply_transform() implementations *must* chain up, so that the
various transformations of each class are preserved. The default
implementation inside ClutterActor applies all the transformations
defined by the scenegraph-related accessors.
Actors performing transformations inside the paint() function will
continue to work as previously.
2009-07-01 12:59:13 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_transformation_matrix:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @matrix: (out): the return location for a #CoglMatrix
|
|
|
|
*
|
|
|
|
* Retrieves the transformations applied to @self
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_get_transformation_matrix (ClutterActor *self,
|
|
|
|
CoglMatrix *matrix)
|
|
|
|
{
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
|
|
|
|
|
|
|
CLUTTER_ACTOR_GET_CLASS (self)->apply_transform (self, matrix);
|
|
|
|
}
|
2009-07-13 15:18:27 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_is_in_clone_paint:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Checks whether @self is being currently painted by a #ClutterClone
|
|
|
|
*
|
|
|
|
* This function is useful only inside the ::paint virtual function
|
|
|
|
* implementations or within handlers for the #ClutterActor::paint
|
|
|
|
* signal
|
|
|
|
*
|
|
|
|
* This function should not be used by applications
|
|
|
|
*
|
|
|
|
* Return value: %TRUE if the #ClutterActor is currently being painted
|
|
|
|
* by a #ClutterClone, and %FALSE otherwise
|
|
|
|
*
|
|
|
|
* Since: 1.0
|
|
|
|
*/
|
|
|
|
gboolean
|
|
|
|
clutter_actor_is_in_clone_paint (ClutterActor *self)
|
|
|
|
{
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self), FALSE);
|
|
|
|
|
|
|
|
/* XXX - keep in sync with the overrides set by ClutterClone:
|
|
|
|
*
|
|
|
|
* - opacity_parent != NULL
|
|
|
|
* - enable_model_view_transform == FALSE
|
|
|
|
*/
|
|
|
|
|
|
|
|
return self->priv->opacity_parent != NULL &&
|
|
|
|
!self->priv->enable_model_view_transform;
|
|
|
|
}
|
2009-11-06 16:55:46 +00:00
|
|
|
|
|
|
|
static void
|
|
|
|
set_direction_recursive (ClutterActor *actor,
|
|
|
|
gpointer user_data)
|
|
|
|
{
|
|
|
|
ClutterTextDirection text_dir = GPOINTER_TO_INT (user_data);
|
|
|
|
|
|
|
|
clutter_actor_set_text_direction (actor, text_dir);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_set_text_direction:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
* @text_dir: the text direction for @self
|
|
|
|
*
|
|
|
|
* Sets the #ClutterTextDirection for an actor
|
|
|
|
*
|
|
|
|
* The passed text direction must not be %CLUTTER_TEXT_DIRECTION_DEFAULT
|
|
|
|
*
|
|
|
|
* If @self implements #ClutterContainer then this function will recurse
|
|
|
|
* inside all the children of @self (including the internal ones).
|
|
|
|
*
|
|
|
|
* Composite actors not implementing #ClutterContainer, or actors requiring
|
|
|
|
* special handling when the text direction changes, should connect to
|
|
|
|
* the #GObject::notify signal for the #ClutterActor:text-direction property
|
|
|
|
*
|
|
|
|
* Since: 1.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_set_text_direction (ClutterActor *self,
|
|
|
|
ClutterTextDirection text_dir)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_if_fail (CLUTTER_IS_ACTOR (self));
|
2009-12-01 16:06:57 +00:00
|
|
|
g_return_if_fail (text_dir != CLUTTER_TEXT_DIRECTION_DEFAULT);
|
2009-11-06 16:55:46 +00:00
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
|
|
|
if (priv->text_direction != text_dir)
|
|
|
|
{
|
|
|
|
priv->text_direction = text_dir;
|
|
|
|
|
|
|
|
/* we need to emit the notify::text-direction first, so that
|
|
|
|
* the sub-classes can catch that and do specific handling of
|
|
|
|
* the text direction; see clutter_text_direction_changed_cb()
|
|
|
|
* inside clutter-text.c
|
|
|
|
*/
|
|
|
|
g_object_notify (G_OBJECT (self), "text-direction");
|
|
|
|
|
|
|
|
/* if this is a container we need to recurse */
|
|
|
|
if (CLUTTER_IS_CONTAINER (self))
|
|
|
|
{
|
|
|
|
ClutterContainer *container = CLUTTER_CONTAINER (self);
|
|
|
|
|
|
|
|
clutter_container_foreach_with_internals (container,
|
|
|
|
set_direction_recursive,
|
|
|
|
GINT_TO_POINTER (text_dir));
|
|
|
|
}
|
|
|
|
|
|
|
|
clutter_actor_queue_relayout (self);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_get_text_direction:
|
|
|
|
* @self: a #ClutterActor
|
|
|
|
*
|
|
|
|
* Retrieves the value set using clutter_actor_set_text_direction()
|
|
|
|
*
|
|
|
|
* If no text direction has been previously set, the default text
|
2009-12-07 19:00:55 +00:00
|
|
|
* direction, as returned by clutter_get_default_text_direction(), will
|
|
|
|
* be returned instead
|
2009-11-06 16:55:46 +00:00
|
|
|
*
|
|
|
|
* Return value: the #ClutterTextDirection for the actor
|
|
|
|
*
|
|
|
|
* Since: 1.2
|
|
|
|
*/
|
|
|
|
ClutterTextDirection
|
|
|
|
clutter_actor_get_text_direction (ClutterActor *self)
|
|
|
|
{
|
|
|
|
ClutterActorPrivate *priv;
|
|
|
|
|
|
|
|
g_return_val_if_fail (CLUTTER_IS_ACTOR (self),
|
|
|
|
CLUTTER_TEXT_DIRECTION_LTR);
|
|
|
|
|
|
|
|
priv = self->priv;
|
|
|
|
|
2009-11-30 16:14:09 +00:00
|
|
|
/* if no direction has been set yet use the default */
|
2009-11-06 16:55:46 +00:00
|
|
|
if (priv->text_direction == CLUTTER_TEXT_DIRECTION_DEFAULT)
|
2009-11-30 16:14:09 +00:00
|
|
|
priv->text_direction = clutter_get_default_text_direction ();
|
2009-11-06 16:55:46 +00:00
|
|
|
|
|
|
|
return priv->text_direction;
|
|
|
|
}
|
actor: Add internal child flag
ClutterActor checks, when destroying and reparenting, if the parent
actor implements the Container interface, and automatically calls the
remove() method to perform a clean removal.
Actors implementing Container, though, might have internal children;
that is, children that are not added through the Container API. It is
already possible to iterate through them using the Container API to
avoid breaking invariants - but calling clutter_actor_destroy() on
these children (even from the Container implementation, and thus outside
of Clutter's control) will either lead to leaks or to segmentation
faults.
Clutter needs a way to distinguish a clutter_actor_set_parent() done on
an internal child from one done on a "public" child; for this reason, a
push/pop pair of functions should be available to Actor implementations
to mark the section where they wish to add internal children:
➔ clutter_actor_push_internal ();
...
clutter_actor_set_parent (child1, parent);
clutter_actor_set_parent (child2, parent);
...
➔ clutter_actor_pop_internal ();
The set_parent() call will automatically set the newly added
INTERNAL_CHILD private flag on each child, and both
clutter_actor_destroy() and clutter_actor_unparent() will check for the
flag before deciding whether to call the Container's remove method.
2009-12-18 23:20:04 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_push_internal:
|
|
|
|
*
|
|
|
|
* Should be used by actors implementing the #ClutterContainer and with
|
|
|
|
* internal children added through clutter_actor_set_parent(), for instance:
|
|
|
|
*
|
|
|
|
* |[
|
|
|
|
* static void
|
|
|
|
* my_actor_init (MyActor *self)
|
|
|
|
* {
|
|
|
|
* self->priv = SELF_ACTOR_GET_PRIVATE (self);
|
|
|
|
*
|
|
|
|
* clutter_actor_push_internal ();
|
|
|
|
*
|
|
|
|
* /* calling clutter_actor_set_parent() now will result in
|
|
|
|
* * the internal flag being set on a child of MyActor
|
|
|
|
* */
|
|
|
|
*
|
|
|
|
* /* internal child: a background texture */
|
|
|
|
* self->priv->background_tex = clutter_texture_new ();
|
|
|
|
* clutter_actor_set_parent (self->priv->background_tex,
|
|
|
|
* CLUTTER_ACTOR (self));
|
|
|
|
*
|
|
|
|
* /* internal child: a label */
|
|
|
|
* self->priv->label = clutter_text_new ();
|
|
|
|
* clutter_actor_set_parent (self->priv->label,
|
|
|
|
* CLUTTER_ACTOR (self));
|
|
|
|
*
|
|
|
|
* clutter_actor_pop_internal ();
|
|
|
|
*
|
|
|
|
* /* calling clutter_actor_set_parent() now will not result in
|
|
|
|
* * the internal flag being set on a child of MyActor
|
|
|
|
* */
|
|
|
|
* }
|
|
|
|
* ]|
|
|
|
|
*
|
|
|
|
* This function will be used by Clutter to toggle an "internal child"
|
|
|
|
* flag whenever clutter_actor_set_parent() is called; internal children
|
|
|
|
* are handled differently by Clutter, specifically when destroying their
|
|
|
|
* parent.
|
|
|
|
*
|
|
|
|
* Call clutter_actor_pop_internal() when you finished adding internal
|
|
|
|
* children.
|
|
|
|
*
|
|
|
|
* Nested calls to clutter_actor_push_internal() are allowed, but each
|
|
|
|
* one must by followed by a clutter_actor_pop_internal() call.
|
|
|
|
*
|
|
|
|
* Since: 1.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_push_internal (void)
|
|
|
|
{
|
|
|
|
ClutterMainContext *ctx = _clutter_context_get_default ();
|
|
|
|
|
|
|
|
ctx->internal_child += 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* clutter_actor_pop_internal:
|
|
|
|
*
|
|
|
|
* Disables the effects of clutter_actor_pop_internal()
|
|
|
|
*
|
|
|
|
* Since: 1.2
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
clutter_actor_pop_internal (void)
|
|
|
|
{
|
|
|
|
ClutterMainContext *ctx = _clutter_context_get_default ();
|
|
|
|
|
|
|
|
if (ctx->internal_child == 0)
|
|
|
|
{
|
|
|
|
g_warning ("Mismatched %s: you need to call "
|
|
|
|
"clutter_actor_push_composite() at least once before "
|
|
|
|
"calling this function", G_STRFUNC);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
ctx->internal_child -= 1;
|
|
|
|
}
|