- We want to get window size from compositor widget as it matches window size used by parent RenderCompositorSWGL rendrer.
- For main thread rendering mCompositorWidget is not available so get window size directly from nsWindow.
Depends on D169906
Differential Revision: https://phabricator.services.mozilla.com/D169907
- Get widget size in RenderCompositorSWGL::BeginFrame() and save it for further use. That prevents flickering/artifacts if compositor widget size is changed during rendering.
- Request full screen update when widget size changes on Wayland.
Differential Revision: https://phabricator.services.mozilla.com/D169905
Remove code on bind/unbind that requested a restyle on an unstyled
element, and that canceled that on an unbound element.
Instead, deal with detached nodes in EffectCompositor.
Tweak test restyles since we now expect one restyle as a result of the
initial composing of the animation that before happened eagerly.
Drive-by remove an unused test function in wpt (animationStartsRightNow
is not defined there anyways).
In practice, this makes it consistent with how we handle bind on
elements in shadow trees.
Differential Revision: https://phabricator.services.mozilla.com/D169932
Bug 1805664 had the side effect of installing terminfo libraries on the
toolchain docker image, which led to clang gaining a dependency on
libtinfo, leading to multiple failures, like PGO run jobs failing with
llvm-profdata failing to load the new dependency, or tsan tests failing
because llvm-symbolizer doesn't work because of the new dependency not
being fulfilled, which in turn breaks TSAN suppressions.
Differential Revision: https://phabricator.services.mozilla.com/D170417
The Grid objects will be regenerated whenever the grid is reflowed, which
is nearly the same to the old check of comparing the contents of the
grid structure. This approach is prone to false positives, which is
explained in a new comment.
Differential Revision: https://phabricator.services.mozilla.com/D169725
This ensures that repeated calls to Element::GetGridFragments will return
an array of idempotent Grid objects for each fragment. This is
accomplished by making the Grid object hold a WeakFrame back to its
originating frame, and updating a property on construction and
destruction.
Differential Revision: https://phabricator.services.mozilla.com/D169724
This patch exposes an unrelated issue that causes a performance regression.
For now, we'll revert this to get back to a normal baseline. Then, separately
fix the underlying code which regresses perf, then re-land this patch.
Revert "Bug 1811978 - Enable the new tiled rendering path, update test expectations r=gfx-reviewers,nical"
This reverts commit 7f3a2568aabf9fa2358fe0f7421042ba85a23442.
Differential Revision: https://phabricator.services.mozilla.com/D170399
For now, the debugger frontend either directly query sourceMapLoader.getGeneratedLocation
or use the util's module helper.
We should unify before the utils module helper so that we can later append
the source object on the output of the sourceMapLoader.
Also simplified a bit various code passing both getState+sourceMapLoader
by handing over the thunkArgs.
The util's module helper requires state, so that this changes forces to
pass thunkArgs to some nested callsites of getGeneratedLocation.
Differential Revision: https://phabricator.services.mozilla.com/D169328
We're going to (ab)use the configure scripts to get the right SDK
locations for msix repackaging, and it would be preferable to avoid
requiring the preprocessor for that.
Differential Revision: https://phabricator.services.mozilla.com/D170171
Rectangle with large floating-point values can start to lose mantissa bits when transformed directly
by the vertex shader, pushed through the viewport transform, and then clipped later. OTOH, the path
fallback (WGR/AA-Stroke or Skia) does not stress floating-point precision as much, so generally
returns more correct results in these cases. This becomes noticeable when the same rectangle is
filled and then subsequently stroked, since the fill may use the rect shader, while the the stroke
will fall back to using a path.
To avoid hitting this issue, this checks if a rect's coordinate are outside as certain reasonable
limit, and if so, falls back to relying on path geometry to handle transform and clipping safely.
If within the limit, the shader precision loss doesn't noticeably impact the results so it is still
safe to use the fast-path.
Differential Revision: https://phabricator.services.mozilla.com/D170270
Recent WebRTC backports and changes that are about to be backported from
upstream to Firefox breaks and will break how we work with PipWire based
desktop capturer. Currently when constructing device list, a fallback to
ScreenCapturerX11 is used, as we don't call set_allow_pipewire(), which
wouldn't make a difference anyway. In such case the only thing we need
is a placeholder for a screen that will request OS level prompt. We also
need a way to request both screens and windows in one xdg-desktop-portal
call as recent WebRTC made each type be called separately, therefore the
introduction of GenericCapturer. Lastly we need to make sure when there
is a MediaDevice requesting the OS prompt, that it will be checked as
first.
In order to use unmodified libwebrtc, Firefox would need to rework the
OS picker to request each type (screens and windows) separately so we
can just use regular ScreenCapturer and WindowCapturer. This should be
done ideally the way Chromium does it, where users can actually see
even the preview of what they picked over xdg-desktop-portal before it
is actually shared with requesting web page and they also have option
to make the request again in case they picked a wrong window or screen.
Differential Revision: https://phabricator.services.mozilla.com/D169627
