Commit Graph

4632 Commits

Author SHA1 Message Date
Emmanuele Bassi
83060b30d1 docs: Remove mentions of "frames" in Timeline
We still have some really old naming lying around.
2011-09-02 13:54:57 +01:00
Emmanuele Bassi
63a05fca9d Lock the main context when modifying the repaint functions list
The repaint functions list can (and should) be manipulated from
different threads, but it currently doesn't prevent multiple threads
from accessing it concurrently. We should have a simple lock and take it
when adding and removing elements from the list; the invocation is still
performed under the Big Clutter Lock™, so it doesn't require special
handling.
2011-09-01 17:12:46 +01:00
Robert Bragg
0aacbd47b7 actor: make offscreen_redirect prop take flags + default off
Because we have had several reports about significant performance
regressions since we enabled offscreen redirection by default for
handling correct opacity we are now turning this feature off by default.

We feel that clutter should prioritize performance over correctness in
this case. Correct opacity is still possible if required but the
overhead of the numerous offscreen allocations as well as the cost of
many render target switches per-frame seems too high relative the
improvement in quality for many cases.

On reviewing the offscreen_redirect property so we have a way to
disable redirection by default we realized that it makes more sense for
it to take a set of flags instead of an enum so we can potentially
extend the number of things that might result in offscreen redirection.

We removed the ability to say REDIRECT_ALWAYS_FOR_OPACITY, since it
seems that implies you don't trust the implementation of an actor's
has_overlaps() vfunc which doesn't seem right.

The default value if actor::redirect_offscreen is now 0 which
effectively means don't ever redirect the actor offscreen.
2011-08-30 16:20:16 +01:00
Emmanuele Bassi
d847d43f70 actions/drag: Be more reliable when destroying the drag handle
Whenever the drag handle gets destroyed mid-drag we need to cancel any
current drag operation and reset the state of the DragAction.

https://bugzilla.gnome.org/show_bug.cgi?id=657681
2011-08-30 13:33:20 +01:00
Alejandro Piñeiro
f1a7cd7c0f Merge branch 'atkwindow' 2011-08-30 12:26:35 +02:00
Emmanuele Bassi
368c916c7f layout-managers: Take into account the allocations's origin
If an actor using a LayoutManager has attributes like margin or padding
then it'll have to shave them from the available allocation before
passing it to the LayoutManager::allocate() implementation. Layout
managers should, thus, not assume that the origin of the allocation is
in (0, 0), but take into account that the passed ActorBox might have a
different origin.

https://bugzilla.gnome.org/show_bug.cgi?id=649631
2011-08-29 22:43:42 +01:00
Emmanuele Bassi
3314385119 box: Munge the allocation passed to the layout manager
The actor is in charge of providing to the LayoutManager the available
allocation. ClutterBox should not just pass the box it got from its
parent: it should, instead, provide a normalized box, with an origin in
(0, 0) and the available size.

https://bugzilla.gnome.org/show_bug.cgi?id=649631
2011-08-29 22:39:14 +01:00
Emmanuele Bassi
96810e1d4d build: Add -lm to clutter's LIBADD rule
We need to explicitly link against libm when building Clutter, since we
use libm API.

https://bugzilla.gnome.org/show_bug.cgi?id=657529
2011-08-29 22:17:53 +01:00
Robert Bragg
2f18855659 Adds experimental clutter_backend_get_cogl_context API
This adds experimental API to be able to get the CoglContext associated
with the ClutterBackend. The CoglContext is required to use some of the
experimental 2.0 Cogl API.

Note: Since CoglContext is itself experimental API this API should
considered experimental too. This patch introduces a
CLUTTER_ENABLE_EXPERIMENTAL_API #ifdef guard which anyone wanting to use
this API must define so it's explicitly clear to developers that they
are playing with experimental API.

Note: This API is not yet supported on OSX because OSX still uses the
stub Cogl winsys and the Clutter backend doesn't explicitly create a
CoglContext.

Note: even though this is experimental API we still promise that it
wont be changed during a stable release cycle. This means for example
that you can depend on this for the lifetime of the clutter-1.8 stable
release cycle.
2011-08-24 16:24:39 +01:00
Alejandro Piñeiro
7635cb6f17 a11y: We don't need to explicitly manage window events 2011-08-23 17:22:17 +02:00
Alejandro Piñeiro
a9690a4a5e a11y: Using AtkWindow on cally-stage 2011-08-23 17:22:11 +02:00
Emmanuele Bassi
2559fb1d0d actor: Add a comment on the unmapping order
Let's add a comment to avoid reshuffling the unmap() implementation and
end up triggering these bugs:

  http://bugzilla.clutter-project.org/show_bug.cgi?id=2621
  https://bugzilla.gnome.org/show_bug.cgi?id=652036
2011-08-23 10:58:09 +01:00
Emmanuele Bassi
d691c9cbe3 x11/settings: Add Fontconfig/Timestamp key
The XSETTINGS key that matches the :fontconfig-timestamp property on
ClutterSettings.
2011-08-18 15:36:20 +01:00
Emmanuele Bassi
7d4f371ee9 settings: Add ClutterSettings:fontconfig-timestamp
The :fontconfig-timestamp is a write-only property that will get updated
by the underlying platform whenever the fontconfig configuration has
been changed — i.e. when the fontconfig caches should be rebuilt after
the user has installed a new font.
2011-08-18 15:33:39 +01:00
Robert Bragg
397fe1b5b6 text: Implement Scriptable interface for font-description
This makes ClutterText implement the Scriptable interface so that we can
have a custom property parser and setter for the font-description
property. This works by simply passing the string description through
to clutter_text_set_font_name.

Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
2011-08-16 18:06:44 +01:00
Robert Bragg
ccb740f167 actor: Don't update last-paint-volume during picking
Actually this change has two notable effects; firstly we no longer
perform culling during picking and secondly we avoid updating the
last-paint-volume of an actor when picking.

We shouldn't perform culling during picking until clutter-stage.c is
updated to setup the clipping planes appropriately.

Since the last-paint-volume is intended to represent the visible region
of the actor the last time it was painted on screen it doesn't make
sense to update this during off screen pick renders since we are liable
to end up with a last-paint-volume that maps to an actors new position
when we next come to paint for real.

This fixes a bug in gnome-shell with dragging dash icons leaving a
messy trail on the screen.

Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
2011-08-15 18:25:43 +01:00
Robert Bragg
d2763fd8a4 stage-cogl: pass top-left relative swap_region coords
All 2D coordinate spaces in Cogl have their origin at the top-left so we
shouldn't be flipping the coordinates we pass to
cogl_framebuffer_swap_region to be relative to the bottom of the
framebuffer.

This bumps the Cogl version requirement to 1.7.5 since we've had to fix
a bug in the semantics of cogl_framebuffer_swap_region.

Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-08-12 15:42:42 +01:00
Alejo Pacin
694649e50c a11y: Notify ATK_STATE_SHOWING state change
https://bugzilla.gnome.org/show_bug.cgi?id=648633
2011-08-04 12:59:50 +02:00
Neil Roberts
91ace65cae clutter-stage-win32: Fix name cogl_win32_onscreen_set_foreign_window
Since commit 38b67e2884 of Cogl the naming scheme for winsys-specific
API has changed to be cogl_win32_onscreen_* instead of
cogl_onscreen_win32_* so it wouldn't build on Windows.
2011-07-31 18:00:51 +01:00
Kirk A. Baker
01fd673505 osx: Avoid leaks in Stage::realize
clutter_stage_osx_realize() can now be called multiple times without
leaking views and windows.

https://bugzilla.gnome.org/show_bug.cgi?id=655307
2011-07-29 13:58:06 +01:00
Kirk A. Baker
e144378f5a osx: Implement windowDidResignKey
Added isHiding field to _ClutterStageOSX to allow windowDidResignKey
delegate to not order full screen window back whilst the full screen
window was being hidden.  This caused other application windows to be
hidden.  Also added code to keep hidden stage windows from being listed
in the application's Windows menu.

https://bugzilla.gnome.org/show_bug.cgi?id=655311
2011-07-29 13:58:01 +01:00
Kirk A. Baker
6f6b27fac5 osx: Add windowDidChangeScreen delegate
Added -windowDidChangeScreen: delegate to handle condition where moving
host window to a different screen would cause pick errors to be output.
The delegate just causes the stage to be redrawn which re-creates the
pick buffer.

https://bugzilla.gnome.org/show_bug.cgi?id=655306
2011-07-29 13:57:13 +01:00
Emmanuele Bassi
7139ada809 model: Remove an unused variable 2011-07-29 13:17:11 +01:00
Evan Nemerson
a6bd11ac62 introspection: add missing introspection data from Vala bindings
https://bugzilla.gnome.org/show_bug.cgi?id=655387
2011-07-29 13:06:51 +01:00
Emmanuele Bassi
fe66575a2c docs: Update with the newly added Animatable symbol 2011-07-29 11:53:16 +01:00
Emmanuele Bassi
f28c1d2d2a state: Use the Animatable interface
The Animatable interface allows object classes to provide and animate
properties outside of the usual GObject property introspection API.

This change allows ClutterState to defer to the animatable objects the
property introspection and animation, just like ClutterAnimation does.
2011-07-29 11:44:28 +01:00
Emmanuele Bassi
8b03ec9d16 actor: There is no need to implement animate_property()
It's a deprecated virtual function, and its replacement does the right
thing by default.
2011-07-29 11:36:18 +01:00
Emmanuele Bassi
677df44975 animation: Use Animatable.interpolate_value()
And stop using the deprecated animate_property() method.
2011-07-29 11:36:18 +01:00
Emmanuele Bassi
8b861cea8f animatable: Deprecate and replace animate_property()
The animate_property() method of the Animatable interface is far less
than optimal:

  • it has a direct reference to ClutterAnimation;
  • it has an interval decomposed as two values.

These issues tie the Animatable interface with the Animation object,
even though it's neither necessary nor future-proof.

Let's introduce a new method, interpolate_value(), which does not
reference ClutterAnimation and uses a ClutterInterval to express the
initial and final states.
2011-07-29 11:36:18 +01:00
Emmanuele Bassi
473f3817d4 model: Clean up ModelIter vfunc wrappers
All ClutterModelIter virtual functions have a default implementation,
and G_TYPE_INSTANCE_GET_CLASS cannot return NULL unless in case of a
catastrophic event in the type system - which will most likely blow up
any application code way before you could call a ModelIter method.

Thus, the idiom:

  klass = CLUTTER_MODEL_ITER_GET_CLASS (instance);
  if (klass && klass->vfunc)
    klass->vfunc (instance);

is utterly useless complication, and it can be perfectly replaced by:

  CLUTTER_MODEL_ITER_GET_CLASS (instance)->vfunc (instance);

without any loss of safety.
2011-07-29 11:29:51 +01:00
Emmanuele Bassi
e470fd7d82 model: Make sure to emit ::row-changed
Currently, only clutter_model_iter_set_valist() is in charge of emitting
the ClutterModel::row-changed signal. Both the set() and the
set_valist() functions can be called with multiple columns, so we
coalesce the signal emission at the end of the set_valist(), to have a
single ::row-changed emission per change.

The clutter_model_iter_set_value() function is just a thin wrapper
around the set_value() virtual function, but since it's called
internally we cannot add the signal emission there as well, as we'd
break the signal coalescing.

For this reason, we need some code refactoring inside the various set()
variants of ClutterModelIter:

  - we only use the internal va_arg variant for both the set() and
    set_valist() public functions, to avoid multiple type checks;
  - the internal set_valist() calls an internal set_value() method
    which calls the virtual function from the iterator vtable;
  - a new internal emit_row_changed() method is needed to retrieve
    the ClutterModel from the iterator, and emit the signal;

Now, all three variants of the value setter will call an internal
ClutterModelIter::set_value() wrapper, and emit the ::row-changed
signal.

To check that the intended behaviour has been implemented, and it's not
going to be broken, the test suite has grown a new unit which populates
a model and changes a random row.
2011-07-28 15:00:18 +01:00
Alejandro Piñeiro
468f121f59 a11y: Use existing clutter_actor_get_paint_visibility
Cally was initially created with Clutter 0.6 in mind. To check
recursively the visibility of a actor a custom method was added.
Since 0.8.4 clutter_actor_get_pain_visibility provides
the same functionality.

Also removed a dummy method. Lets add methods that provide a real
functionality.
2011-07-26 16:24:49 +02:00
Emmanuele Bassi
7f8838d7cc cairo-texture: Add the :auto-resize property
Keeping the backing Cairo surface of a CairoTexture canvas in sync with
the actor's allocation is tedious and prone to mistakes. We can
definitely do better by simply exposing a property that does the surface
resize and invalidation automagically on ::allocate.
2011-07-26 14:55:19 +01:00
Emmanuele Bassi
0f8e3ab4aa bin-layout: Fix a leak
https://bugzilla.gnome.org/show_bug.cgi?id=655342
2011-07-26 13:55:24 +01:00
Emmanuele Bassi
78049c38bf docs: Fixes for building the API reference 2011-07-26 13:44:12 +01:00
Emmanuele Bassi
278daca61c cairo-texture: Deprecate create()/create_region()
The recommended way of drawing on a ClutterCairoTexture is the ::draw
signal.
2011-07-26 12:53:22 +01:00
Emmanuele Bassi
2f445682b1 cairo-texture: Use signal-based drawing
The current "create context/draw/destroy context" pattern presents
various problems. The first issue is that it defers memory management to
the caller of the create() or create_region() methods, which makes
bookkeeping of the cairo_t* harder for language bindings and third party
libraries. The second issue is that, while it's easier for
draw-and-forget texturs, this API is needlessly complicated for contents
that have to change programmatically - and it introduces constraints
like calling the drawing code explicitly after a surface resize (e.g.
inside an allocate() implementation).

By using a signal-based approach we can make the CairoTexture actor
behave like other actors, and like other libraries using Cairo as their
2D drawing API.

The semantics of the newly-introduced ::draw signal are the same as the
one used by GTK+:

  - the signal is emitted on invalidation;
  - the cairo_t* context is owned by the actor;
  - it is safe to have multiple callbacks attached to the same
    signal, to allow composition;
  - the cairo_t* is already clipped to the invalidated area, so
    that Cairo can discard geometry immediately before we upload
    the texture data.

There are possible future improvements, like coalescing multiple
invalidations inside regions, and performing clipped draws during
the paint cycle; we could even perform clipped redraws if we know the
extent of the invalidated area.
2011-07-26 12:40:52 +01:00
Emmanuele Bassi
e1327f5591 actor: Clean up the allocate() checks
Move the conditions to their own boolean flags, to avoid making the
if() incomprehensible.
2011-07-19 14:03:02 +01:00
Emmanuele Bassi
123bd41e6c texture: Clean up error reporting
ClutterTexture relies too much on GError, even for things that are
clearly programmer errors. Also, no error message passed to GError
is marked for translation as it should.

We should move the programmer errors, like passing the wrong bpp
value with regards to the presence of the alpha channel, to real
warnings; we should also try and harmonize all the error messages,
and not mention Cogl — especially in the ones marked for translation.
2011-07-18 18:06:57 +01:00
Robert Bragg
672bc337ba build: don't explicitly include and egl or gl headers
This avoids explicitly including gl or egl headers in
clutter-egl-headers.h. We were getting build failures when building
clutter against a libcogl that has runtime support for GL and GLES
because cogl-defines.h was including gl.h and then clutter-egl-headers.h
was later including GLES2/gl.h with typedef conflicts. Clutter relies on
Cogl to abstract GL and GLES and the winsys APIs like EGL and GLX so
Clutter should just rely on cogl.h to include the appropriate egl.h in
clutter-egl-headers.h.

Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Reviewed-by: Neil Roberts <neil@linux.intel.com>
2011-07-18 17:07:40 +01:00
Damien Lespiau
a4b9eaf3ec alpha: Don't use the GClosure to compute the alpha when possible
Setting up the call and calling the GClosure was showing up in profiles
and seemed an easy one to remove.

Instead of calling the closure, let's remember the alpha func and the
user_data when possible (ie set_mode() and set_func()) and use it in
get_alpha().

https://bugzilla.gnome.org/show_bug.cgi?id=654727
2011-07-18 15:17:15 +01:00
Damien Lespiau
4d58534fbb text: Fix the default value of "editable"
The "editable" property is documented to default to TRUE, but is
initialized to FALSE in the _init() function.

Third party code would be affected if we changed the default to be
TRUE, so we have to change the default value in the GParamSpec.

https://bugzilla.gnome.org/show_bug.cgi?id=654726
2011-07-18 13:57:44 +01:00
Emmanuele Bassi
e25e48efb3 Do not use return with void functions
https://bugzilla.gnome.org/show_bug.cgi?id=654718
2011-07-18 13:53:16 +01:00
Emmanuele Bassi
1402c05112 timeline: Cast elapsed time before emitting ::new-frame
When emitting a new-frame signal, priv->elapsed_time is passed as a
parameter. This is a gint64. The closure marshal uses an INT. On some
platforms, this is not received correctly by signal handlers (they
receive 0). One solution is to cast priv->elapsed_time to a gint when
emitting the signal.

We cannot change the signature of the signal without breaking ABI.

https://bugzilla.gnome.org/show_bug.cgi?id=654066
2011-07-18 13:45:55 +01:00
Bastian Winkler
c9cbe58343 snap-constraint: Fix allocation for CLUTTER_SNAP_EDGE_RIGHT
Use source_with instead of source_height to calculate the the x1/x2
positions for CLUTTER_SNAP_EDGE_RIGHT

https://bugzilla.gnome.org/show_bug.cgi?id=654024
2011-07-18 13:43:29 +01:00
Emmanuele Bassi
bc1c3e003d x11/event: Avoid compiler warnings with HAVE_XGE undefined 2011-07-15 16:08:38 +01:00
Emmanuele Bassi
538559391c actor: Plug a leak in an error path
If the meta for the animation property is not found, the name of the
property to look for is still from the token, and we need to free the
memory allocated for it.
2011-07-15 15:53:55 +01:00
Carlos Garnacho
075a4ed86c x11: Improve XGenericEventCookie data allocation
https://bugzilla.gnome.org/show_bug.cgi?id=654656

Clutter may be used together with GTK+, which indirectly may use
XInput2 too, so the cookie data must persist when both are handling
events.

What happens now in a nutshell is, Clutter is only guaranteed to allocate
the cookie itself after XNextEvent(), and only frees the cookie if its
XGetEventData() call allocated the cookie data.

The X[Get|Free]EventData() calls happen now in clutter-event-x11.c as
hypothetically different event translators could also handle other set
of X Generic Events, or other libraries handling events for that matter.
2011-07-15 13:46:33 +02:00
Robert Bragg
c6a0427c69 pick: Use Cogl API not GL API to disable dithering
When picking we need to disable dithering to be sure that the hardware
will not modify the colors we use as actor identifiers. Clutter was
manually calling glEnable/Disable GL_DITHER to handle this, but that was
a layering violation since Cogl is intended to handle all interactions
with OpenGL. Since we are now striving for GL vs GLES to be a runtime
choice we need to remove this last direct usage of GL from Clutter so it
doesn't have to be linked with GL at build time.

Signed-off-by: Neil Roberts <neil@linux.intel.com>
2011-07-14 13:54:16 +01:00
Robert Bragg
3183240fef Provide stable paint-boxes for fixed sized actors
This updates _clutter_paint_volume_get_stage_paint_box to try and
calculate more stable paint-box sizes for fixed sized paint-volumes by
not basing the size on the volume's sub-pixel position.

So the aim is that for a given rectangle defined with floating point
coordinates we want to determine a stable quantized size in pixels that
doesn't vary due to the original box's sub-pixel position.

The reason this is important is because effects will use this API to
determine the size of offscreen framebuffers and so for a fixed-size
object that may be animated across the screen we want to make sure that
the stage paint-box has an equally stable size so that effects aren't
made to continuously re-allocate a corresponding fbo.

The other thing we consider is that the calculation of this box is
subject to floating point precision issues that might be slightly
different to the precision issues involved with actually painting the
actor, which might result in painting slightly leaking outside the
user's calculated paint-volume. This patch now adds padding to consider
this too.

Signed-off-by: Neil Roberts <neil@linux.intel.com>
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
2011-07-14 13:54:12 +01:00