lp://staging/~kruk87/glmark2/wayland-opaque-region
- Get this branch:
- bzr branch lp://staging/~kruk87/glmark2/wayland-opaque-region
Branch merges
- Alexandros Frantzis: Approve
-
Diff: 38 lines (+10/-0)2 files modifiedsrc/native-state-wayland.cpp (+9/-0)
src/native-state-wayland.h (+1/-0)
Related bugs
Bug #1219764: Wayland: the window should be opaque | Undecided | Fix Committed |
Related blueprints
Branch information
Recent revisions
- 281. By Rafal Mielniczuk
-
NativeStateWayland: make wayland window opaque
We need to make window opaque in order to prevent redrawing background
behind it in some of the benchmarks, eg. jellyfish. - 280. By Jesse Barker
-
Fix for bug 1208996. Merged from lp:~jbkonno/+junk/glmark2/revision/279.
Add stub methods for new Wayland output events.
- 277. By Rafal Mielniczuk
-
NativeStateWayland: improve destructor
We must check if objects have been allocated before deleting them. This
prevents crashes when invoking client with eg. --list or --help flags.
This patch also cleanup some unused method declarations from header
file. - 276. By Jesse Barker
-
Scene{Shadow,
Refract} : When testing on Nexus 10, it was discovered that the
default texture allocation for these scenes can blow out the max texture size
for an implementation (canvas size on that platform is 2560 x 1504, which
makes us try to allocate textures at 5120 x 3008). So, clamp the texture
dimensions while maintaining proper aspect ratio. - 275. By Jesse Barker
-
CanvasGeneric: Check buffer object handles before calling delete. While it is
perfectly legal and harmless to delete an object handle of zero, there are
times when we do this before having made a context current, which _should_ be
harmless (because implementations don't like to crash), but it not really cool. - 274. By Jesse Barker
-
SceneCollection: Consolidate the scene vector population into an object to
simplify adding scenes across platforms (rather than having to add for Android
separately from others). - 272. By Jesse Barker
-
SceneDesktop: RenderObject:
:init() was creating a zero-sized texture image and
attaching it to a framebuffer object. This is not illegal, but it does violate
the spec rules for framebuffer completeness. Some implementations are doing
the check immediately at attach time rather than deferring to draw time, which
is not really "correct", but we can live with it. Defer the image
(re)creation to resize time and (re)attach it to the framebuffer object then.
Branch metadata
- Branch format:
- Branch format 7
- Repository format:
- Bazaar repository format 2a (needs bzr 1.16 or later)
- Stacked on:
- lp://staging/glmark2/2011.11