| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When the first child of a box layout container is invisible, the
position of the last displayed child is wrong.
ClutterBoxLayout::allocate() first computes the sizes needed for
each of the visible children and stores those sizes into a 'sizes'
array (the array counter is incremented on visible children only).
Later on, ClutterBoxLayout::allocate() computes the position of
each visible children, using the 'sizes' array. However, it uses the
child index to access the array, instead of a 'visible child index'.
https://bugzilla.gnome.org/show_bug.cgi?id=669291
|
|
|
|
|
|
|
|
|
|
|
|
| |
When handling Configure events from the X server we update the
internal copy of the window size. Unfortunately we may be updating the
wrong stage implementation because we use the one related to the event
translator (which is the first created stage).
This patch fix flickering/redrawning issues with multi-stage by
looking for the right stage implementation associated with an XEvent.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@linux.intel.com>
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If you execute the following sequence :
stage = clutter_stage_new ();
clutter_actor_set_size (stage, 1280, 800);
clutter_actor_realize (stage);
Then you end up when ClutterStageCogl creating an onscreen buffer of
size 1280x800 but ClutterStageX11 storing the stage size at 640x480.
This patch resync the 2 implementation by using the ClutterStage's
size in both classes when realizing.
Signed-off-by: Lionel Landwerlin <llandwerlin@gmail.com>
https://bugzilla.gnome.org/show_bug.cgi?id=667540
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The allocation code for BoxLayout contains a sequence of brain farts
that make it barely working since the synchronization of the layout
algorithm to the one in GtkBox.
The origin of the layout is inverted, and it doesn't take into
consideration a modified allocation origin (for actors the provide
padding or margin).
The pack-start property is broken, and it only works because we walk the
children list backwards; this horribly breaks when a child changes
visibility. Plus, we count invisible children, which leads to
allocations getting insane origins (either close to -MAX_FLOAT or
MAX_FLOAT).
Finally, the allocation is applied twice even for non-animated cases.
https://bugzilla.gnome.org/show_bug.cgi?id=669291
|
|
|
|
| |
This reverts commit 755216cbbecbd36c506c40599787b7d623201436.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Cogl's API has been changed in the 1.9 cycle; specifically the
CoglVector3 type has been removed and replaced by a float[3]. This
introduced an upper bound on the version of Cogl that we can use
when building Clutter.
To remain on the safe side, let's specify that we can only build
Clutter 1.8 with a Cogl version newer than 1.8.0, but older than
1.9.2.
Distro teams will hate us, but that's how we roll.
|
|
|
|
|
|
| |
(cherry picked from commit 0c365f9f4c920274e1c1b8190b6bc804260355c6)
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
|
|
|
|
|
|
| |
Do not try to disconnect signal handlers from NULL pointers.
(cherry picked from commit cce2e25ac7ef395d880352c93272256ea088c2dd)
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
|
|
|
|
|
|
| |
Do not try to disconnect signal handlers from NULL pointers.
(cherry picked from commit dbf3b607bf1b295c782ab8edfb2d80bc25f74a37)
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Sadly, we need to remove the G_GNUC_NULL_TERMINATED annotation from
ClutterBox packing API; the compiler will otherwise emit a warning
for perfectly legal statements like:
clutter_box_pack (box, child, NULL);
because of the missing sentinel.
See also: g_object_new().
(cherry picked from commit 65889fa075f6123caf0d47568413466e039b5eef)
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
|
|
|
|
|
|
|
| |
If a font description is not set to have an absolute size then we were
using the wrong transformation for points to device units.
(cherry picked from commit 158245fda31e7adad38aa49e4c12efcad0e326e0)
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
|
|
|
|
|
|
|
| |
If ClutterSettings:font-dpi is set to -1 then the default resolution
should be returned by ClutterBackend — i.e. 96.0.
(cherry picked from commit 943a65b476431400bd0436e1be243a2917385866)
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
|
|
|
|
| |
Once you add the missing translator comments, it makes a lot
more sense...
|
|
|
|
|
|
| |
(cherry picked from commit dc77a5ef310c8da5a126f90a21a75520b5d925de)
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
|
|
|
|
|
|
|
|
|
|
|
| |
While working through the Python3/pygobject bindings, I came across a missing
(allow-none) in clutter_state_set_key(). This allows the API to specify to None
as the source_target.
https://bugzilla.gnome.org/show_bug.cgi?id=664996
(cherry picked from commit 1e05e8ceaff25dfb3600cf0798fbca23c1725e12)
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
|
|
|
|
|
|
|
|
|
|
| |
Add out annotations for clutter_path_get_node() and
clutter_path_get_position()
https://bugzilla.gnome.org/show_bug.cgi?id=664336
(cherry picked from commit 6336883da7aa15ec06d9f3218ad5c617e57af0a5)
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
|
|
|
|
|
|
|
|
|
| |
ClutterInputDeviceX11 has been made private, so we cannot access it from
outside of clutter-input-device-core-x11.c. We should have simple
accessors for the min/max keycode, which is the only detail that we use.
(cherry picked from commit 1dc7c4543896bfc2f2cd460cdea9c417aefd7952)
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously the offscreen effect was keeping track of the size of the
texture so that it could detect when a different size is requested and
create a new texture. However this breaks if a subclass overrides
create_texture to make the texture bigger because in that case the
size of the texture will always be different from the calculated size
of the actor. This patch makes it also track the size of the fbo that
was requested before being passed through create_texture() and it
instead uses that to detect when a new FBO is needed.
https://bugzilla.gnome.org/show_bug.cgi?id=665040
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
(cherry picked from commit a2774fb0dcce0c92036b69fb75092ec8dc80905d)
|
|
|
|
|
|
|
|
|
| |
If our check of the CoglOnscreenTemplate during initialization fails
then we disable the request for an alpha component in the swap chain and
try the check again.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=661022
(cherry picked from commit 6d974f278e937cce0872567bce672ed515662678)
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
|
|
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=660985
(cherry picked from commit 5dbc00e625c515a6732eaec06ddd6185ad8b5916)
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When the viewport gets expanded because the actor extends off the edge
of the screen, instead of applying the transformation to the root of
the modelview transformation it is now applied to the end of the
projection transformation. This should end up with the same
transformation. This fixes a problem when the offscreen effects are
nested and the inner effect would try to pick up the current modelview
transformation to rescale it to fit the new viewport size. In this
case the modelview would have already been scaled for the size of the
outer viewport so it would end up wrong.
https://bugzilla.gnome.org/show_bug.cgi?id=659601
Reviewed-by: Emmanuele Bassi <ebassi@linux.intel.com>
Reviewed-by: Robert Bragg <robert@linux.intel.com>
(cherry picked from commit c67d3e5b52955fc87b5c1581a82b745d9b2b047c)
Signed-off-by: Emmanuele Bassi <ebassi@linux.intel.com>
|
| |
|
| |
|
|
|
|
|
|
|
| |
If we do project() → get_bounding_box(), we'll try to complete the
volume twice, which whacks out all the lazily computed vertices.
Reviewed-by: Robert Bragg <robert@linux.intel.com>
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
This makes it much clearer, and fixes some typos in the original.
https://bugzilla.gnome.org/show_bug.cgi?id=659893
|
| |
|
| |
|
| |
|
| |
|
| |
|