The fundamental issue here is that `ParentCrashTest` starts a second
`GeckoRuntime` via `RemoteGeckoService` within the same logical Android
Application.
This patch adds `RuntimeCreator.shutdownRuntime()` so that the test can
shut down the existing `GeckoRuntime` before starting up its own.
Differential Revision: https://phabricator.services.mozilla.com/D61249
--HG--
extra : moz-landing-system : lando
Every build is a clobber, so we don't need to clean things up at the beginning
of a build.
Differential Revision: https://phabricator.services.mozilla.com/D61608
--HG--
extra : moz-landing-system : lando
Mozharness currently deals with buildid in two places:
- It passes `MOZ_BUILD_DATE` through the environment in a complicated way,
and sets it if it isn't set.
- It verifies that the buildid of the generated package matches that the one it
determined.
The former is duplicate code, and the later requires knowing details of the
build for little value.
Differential Revision: https://phabricator.services.mozilla.com/D61607
--HG--
extra : moz-landing-system : lando
I'm guessing that maybe this was a leftover from when we built arm64-windows with MSVC. In any case, the plugin seems to work fine now.
Differential Revision: https://phabricator.services.mozilla.com/D61620
--HG--
extra : moz-landing-system : lando
Seems they were only used for the old XBL marquee implementation and have no
remaining callers.
Differential Revision: https://phabricator.services.mozilla.com/D61338
--HG--
extra : moz-landing-system : lando
When our position changes from / to absolute / fixed, then we need to reframe
(most likely, at least). But in some easier cases we can just live with a reflow
plus (optionally) a containing-block update or a repaint, depending on the case.
We need to delete the normal position property when this happens and we switch
from static to relative/sticky, and also need to update the handling of
UpdateContainingBlock to avoid making decisions based on position not changing.
This will actually not cause more reframes, as one would expect, because that
optimization became somewhat obsolete by bug 1519371 (which made the
optimization exact).
Differential Revision: https://phabricator.services.mozilla.com/D61142
--HG--
extra : moz-landing-system : lando
No other items need to round their bounds/clips, and backdrop filters
should be no exception. They should trust scene building (and to a
lesser extent, frame building) to perform any necessary snapping. This
patch also removes the ToRoundedLayoutPoint/Rect methods as there are no
more users of the APIs, nor should there ever be.
Differential Revision: https://phabricator.services.mozilla.com/D61618
--HG--
extra : moz-landing-system : lando
Calling gtk_window_hide without pausing renderer cause crashes. Also to avoid repeated calls of NativeMoveResizeWaylandPopup from NativeMoveResizeWaylandPopup we don't call Resize or Move anymore, we just update mBounds which is enough.
Differential Revision: https://phabricator.services.mozilla.com/D61312
--HG--
extra : moz-landing-system : lando
Webrender positions sticky items relative to the viewport rect set on
their ancestor scroll frame, which it expects to be in local
space. The viewport should be set to what gecko calls the scroll
frame's "scroll port" (relative to its reference frame).
Previously we were setting this to be `composition bounds /
resolution`, which gives the correct results at the initial zoom
level, but causes sticky items to jump around when the zoom
changes. This occurs because the composition bounds of the RCD-RSF do
not change as the zoom changes, and instead remain equal to the
widget's size.
The layout viewport remains at the correct size when zoomed, but
unfortunately doesn't have the correct origin. And FrameMetrics does
not provide the constant "initial zoom" value we require to obtain the
desired viewport rect from the composition bounds. To fix this we
therefore must query the scroll port and offset directly from the
scrollable frame, meaning the value remains correct even as the zoom
changes.
Differential Revision: https://phabricator.services.mozilla.com/D60970
--HG--
extra : moz-landing-system : lando
This is done for both inline and editor grid. This gives us a
clearer way the grid should be used, and also allows us to
not have dedicated rules for editor layout in some cases.
Differential Revision: https://phabricator.services.mozilla.com/D61567
--HG--
extra : moz-landing-system : lando
The devtools tests are already unconditionally disabled (bug 1467313),
so there is no loss of coverage by removing the in-process test
scenario.
The other two tests already pass without disabling remote extensions.
Differential Revision: https://phabricator.services.mozilla.com/D60974
--HG--
extra : moz-landing-system : lando
Breaks the dependency from Selection.h. Allows reuse for Selection
across the Shadow DOM boundary, which will require styled ranges to be
stored in `DocumentOrShadowRoot` too.
Differential Revision: https://phabricator.services.mozilla.com/D61549
--HG--
extra : moz-landing-system : lando
There is no guarantee that a tab has finished loading when the test
resumes. As a result, it is possible for the page action's visibility
state to reset (to invisible) between calling `pageAction.show()` and
`pageAction.isShown()`. To prevent that from happening, wait until the
tab has finished loading before resuming the test.
Differential Revision: https://phabricator.services.mozilla.com/D61125
--HG--
extra : moz-landing-system : lando
Run the tests from 70490abb43 locally.
BaseURI is accessed via parent when possible so that we handle the very unlikely case that some chrome code
uses Attr to access svg:use's clone's baseURI via Attr.
Differential Revision: https://phabricator.services.mozilla.com/D61507
--HG--
extra : moz-landing-system : lando