of overflowing the id numbers when continusly creating and destroying
actors on long running applications for 16bpp.
* clutter/clutter-private.h: replaced hashtable with GArray and a
GSList for available slots in the array.
* clutter/clutter-actor.c: (create_actor_id): function to create an
actor->id mapping, (release_actor_id): function to mark an existing id
as available for reuse.
* clutter/clutter-main.c: (clutter_context_free): added utility
function for cleaning up the context,
(clutter_get_actor_by_gid): use the GArray for looking up actors.
* clutter/clutter-actor.c: Remove the usage of g_return_if_fail()
from static functions: either assert or use g_warning() to check
internal state, as g_return_if_fail() can be compiled out.
* clutter/clutter-main.c: Ditto as above.
* clutter/x11/clutter-backend-x11.h:
* clutter/x11/clutter-event-x11.c:
* clutter/x11/clutter-x11.h:
(clutter_x11_handle_event):
(clutter_x11_disable_event_retrieval):
Functions to allow to hook into external XEvent retrieval (for
example when using clutter with gtk); NB: this API is tentative.
* clutter/x11/clutter-event-x11.c (event_translate): Ignore
PropertyNotify events if they don't match our window.
* clutter/x11/clutter-stage-x11.c:
(clutter_stage_x11_request_coords): Chain up to the parent's
request coords.
* clutter/x11/clutter-event-x11.c (event_translate): Ignore
ConfigureNotify events if we are painting on a foreign window.
* clutter/x11/clutter-stage-x11.c:
(clutter_stage_x11_request_coords): Reliquish control of the
window size, if we are painting on a foreign window. This fixes
the GtkClutterEmbed widget.
* clutter/x11/clutter-x11.h:
* clutter/x11/clutter-backend-x11.c:
(clutter_x11_set_display):
Function to set X display connection prior to calling
clutter_init(); stripped loads of trailing space.
* clutter/clutter-actor.h:
* clutter/clutter-actor.c:
(clutter_actor_get_abs_opacity): Add function that does what
get_opacity() does now...
(clutter_actor_get_opacity): ... and make get_opacity() do what
it's supposed to be doing. The original get_opacity() returned
a composited value, and there's no way to actually extract the
real opacity value set with set_opacity().
* clutter/clutter-clone-texture.c:
* clutter/clutter-rectangle.c:
* clutter/clutter-texture.c: Update to use get_abs_opacity().
* clutter/clutter-entry.c:
* clutter/clutter-label.c: Ditto. Also, never change the stored
alpha value. (#804)
* tests/Makefile.am:
* tests/test-opacity.c: Test suite for the get_opacity() and
get_abs_opacity() API, and correct opacity handling.
* README: Add note about the change in get_opacity().
* clutter/clutter-actor.[ch]:
* clutter/clutter-types.h:
* doc/reference/clutter-docs.sgml: Fix a lot of documentation.
2008-02-15 Matthew Allum <mallum@openedhand.com>
* clutter/clutter-actor.c:
* clutter/clutter-actor.h:
(clutter_group_query_coords):
Added function to translate vertex array to a bounding box.
(clutter_actor_is_scaled):
(clutter_actor_is_rotated):
Fixed documentation.
* clutter/clutter-actor.c:
* clutter/clutter-actor.h:
(clutter_actor_is_scaled):
(clutter_actor_is_rotated):
Convenience functions to test whether actor is scaled or rotated.
(clutter_actor_apply_relative_transform_to_point):
Removed unused variable.
* clutter/clutter-group.c:
(clutter_group_query_coords):
Use clutter_actor_get_relative_vertices() to calculate bounding
boxes of children that are scaled or rotated.
* clutter/clutter-actor.c:
* clutter/clutter-actor.h:
(clutter_actor_get_relative_vertices):
(clutter_actor_apply_relative_transform_to_point):
Functions to calculate actor vertices in the plane of a given
ancestor.
* clutter/clutter-texture.c (texture_update_data):
Fixed incorrect calculation of offset into texture data (needs to
use width of the texture, not of the updated area).
* clutter/clutter-actor.c:
Minor documentation tweak to class description.
* clutter/clutter-behaviour-scale.c:
'Force' start + end vals of scale behaviour
(#779, Havoc Pennington)
* clutter/clutter-actor.c (clutter_actor_transform_stage_point):
Moved part of the calculation to floating point to avoid overflow
of ClutterFixed range for large actors (bug 613).
* clutter/clutter-behaviour-path.c: Move the nice graph and
description from the Path behaviour...
* clutter/clutter-behaviour.c: ... to the main Behaviour class.
* clutter/clutter-behaviour-path.c: Reference the newly
added image showing the effects of different alpha functions
on the same path behaviour.
* configure.ac: Find out the prefix of some of the libraries
we depend upon, so that gtk-doc can fix the references for us.
* clutter/clutter-script.c:
(resolve_alpha_func): Use a whitelist containing the alpha function
provided by Clutter, as a fallback in case g_module_open() does not
work.
(clutter_script_parse_alpha): Add an explicit warning in case we
cannot find the alpha function used inside a UI definition.
implementation of a box-blur shader from Gwenole Beauchesne that
should work on GLSL implementing low-end IGPs not implementing
dynamic branching. The original code used crashed some of these
due to bugs in drivers or similar. Resolves bug #710.
* clutter/clutter-actor.c:
(clutter_actor_set_property):
Fixed handling of PROP_ROTATE_CENTER_Y, ROTATE_CENTER_Z
* tests/test-actors.c:
Adjusted to use anchor point for both group and hands.
* tests/test-actors.c: (frame_cb), (main):
Changed set_anchor to move_anchor.
Changed clutter_actor_rotate x & y values to 0,0.
Switched on scaling because it seems to work without drifting.
* clutter/clutter-script-parser.c:
(clutter_script_get_type_from_symbol),
(clutter_script_get_type_from_class): Use BIND_LAZY flag
when looking at the symbols, so we don't load them all up.
* clutter/clutter-script.c (resolve_alpha_func): Ditto as above.
(_clutter_actor_apply_modelview_transform): perform translation as
well as anchor point adjustment based on self not being stage (and
not by self not being an orphan).
* clutter/clutter-texture.c:
Initial attempt at handling resizing/tranformed source textures.
* clutter/pango/pangoclutter-render.c: (draw_glyph):
Guards against freetype returning NULL bitmaps and
thus causing device by zeros on NVIDIA drivers. (see ml)
* clutter/clutter-actor.c: (clutter_actor_get_vertices):
Make sure GL related Matrices are initialised. Fixes#756
* clutter/clutter-texture.c:
Fix up code from Norways answer to Jean-Michel Basquiat.
(warning on tile_get_pixbuf)
renamed clutter_actor_apply_shader to be clutter_actor_set_shader.
Reindented/aligned header.
* clutter/clutter-texture.c: (clutter_texture_paint): temporarily
disable the shader on top of the clutter context's shader stack while
rendering into the FBO.
* tests/test-fbo.c: (main): s/apply_shader/set_shader/
* tests/test-shader.c: (button_release_cb), (main):
s/apply_shader/set_shader/
* clutter/clutter-shader.c:
Minor formatting cleanups to fit in 80 cols.
* clutter/clutter-texture.c:
More safety checks, clean ups in clutter_texture_new_from_actor()
* clutter/cogl/gl/cogl.c:
Always clear the FBO initially when rendering
* tests/test-fbo.c:
Overhall the test as to be more useful (and show current issues)
* clutter/cogl/cogl.h:
* clutter/cogl/gl/cogl.c:
* clutter/cogl/gles/cogl.c:
Disable the depth test and buffer as has some strange side
effects, mainly on x/y axis rotation with multiple layers at
same depth (eg rotating text on a bg has very strange
effect). Seems no clean 100% effective way to fix without other
odd issues.. So for now move to application to handle and add
cogl_enable_depth_test() as for custom actors (i.e groups) to
enable if need be.
* clutter/clutter-feature.h:
* clutter/clutter-texture.c:
* clutter/clutter-texture.h:
* clutter/cogl/cogl.h:
* clutter/cogl/gl/cogl.c:
* clutter/cogl/gles/cogl.c:
* tests/Makefile.am:
* tests/test.fbo.c:
Add initial support for FBO's in Clutter (OpenGL only so far).
See new clutter_texture_new_from_actor()
Initial implementation, needs work.
* clutter/x11/clutter-stage-x11.c:
(clutter_stage_x11_set_cursor_visible):
Fall back to again not relying on xfixes to hide cursor. *sigh*
* clutter/clutter-deprecated.h:
Add clutter_group_find_child_by_id
* clutter/clutter-actor.c:
(clutter_actor_get_property): Use clutter_actor_get_reactive()
instead of the IS_REACTIVE() macro.
(clutter_actor_set_reactive): Emit the notify signal for the
reactive property.
* clutter/clutter-timeline.c:
(timeline_timeout_func),
(clutter_timeline_get_delta): Coding style fixes.
* clutter/clutter-timeline.c (clutter_timeline_skip): Don't
add the frames outside the direction check. (thanks to benoar
on clutter-list for spotting it)
* clutter/clutter-actor.c:
(clutter_actor_transform_vertices):
Call clutter_actor_query_coords() instead of accessing the actor
box, so that actors that calculate their bounding box on the fly
(like ClutterGroup) work.
* (_clutter_actor_apply_modelview_transform_recursive):
Made to work for parentless actors.
clutter_timeline_set_direction instead of static state.
(raise_top): made rectangle remain at fixed depth and raise
above each other.
(janus_group): added a rectangle with two faces, one red and
one green, that spins around.
* clutter/clutter-entry.h:
* clutter/clutter-entry.c:
(clutter_entry_set_cursor_position),
(clutter_entry_get_cursor_position): Rename [gs]et_position()
to [gs]set_cursor_position(), and avoid method name collisions
with bindings for high-level languages.
* clutter/clutter-deprecated.h: Add replacement warnings for the
renamed methods.
* clutter/clutter-actor.c (clutter_actor_init): Explicitly
memset() to 0 the clip container, so if you get the clip without
having set it, you won't get garbage.
2008-01-31 Rob Bradford <rob@openedhand.com>
* README: Add the osx flavour, and mark it experimental,
along with the sdl backend. EGL-based backends should not be
considered experimental anymore, even though they might require
tweaking.
* HACKING: Expand the "document API" point, and the release
process.
* README: Update the release notes regarding the scale behaviour,
now that the gravity has been removed.
* clutter/clutter-label.c (clutter_label_query_coords): Check
if the layout logical rectangle size is bigger than the bounding
box; this usually means that the bounding box changed in one
point and the label private allocation needs to be
updated. (#724)
