We are downgrading this assertion since it's clearly happening and we
want a patch that is upliftable to beta; we will investigate and
address the potential logic problem separately. (It could also just
be flipped memory bits; we need to investigate.)
Differential Revision: https://phabricator.services.mozilla.com/D206027
This patch reverts the changes of Bug 1884000 which was trying to
address an intermittent failure. It did not seem to help so there is no
real reason to keep these changes.
Differential Revision: https://phabricator.services.mozilla.com/D205883
This changes the resize check to an objective standard of whether or not
the widget has resized, rather than asking the widget if it is in the
process of resizing.
Differential Revision: https://phabricator.services.mozilla.com/D202869
FlushRendringAsync was added in bug 1361257 in order to not block the main thread when doing some widget interactions on windows. To that end it made the IPDL message async without changing the behavior of the code on the compositor thread. Since the widget just needs to tell the compositor to refresh without needing to block on it, it does not require the compositor thread itself to block until rendering is done, so this patch makes it so FlushRenderingAsync no longer has synchronous behavior on the compositor thread. This helps whith avoiding UI freezes when FlushRenderingAsync is send while an abnormally large workload is being processed by the low priority scene builder thread and workers.
Differential Revision: https://phabricator.services.mozilla.com/D202921
Git isn't able to detect the file being only renamed in the previous changeset
as the changes made to it are too important.
So, use a distinct changeset to apply the file rename.
Differential Revision: https://phabricator.services.mozilla.com/D205988
Change build-apk, build-bundle, test-components, and test-apk to b-linux-medium-gcp, to get the same machine type used traditionally on github. Change ui-test-apk to t-linux-large, similar but smaller than what was used on github.
Differential Revision: https://phabricator.services.mozilla.com/D205937
The source-host-path option can be set to let mach vendor find a python
module/class next to moz.yaml, that is then used as source host.
Differential Revision: https://phabricator.services.mozilla.com/D205710
Remote canvas can run in the GPU process, and if the GPU process
crashes, we need to notify the application using canvas. Historically we
just failed, and the application may have been able to continue drawing
but with the contents prior to the crash lost. Later we regressed to
prevent the canvas from being used at all.
This patch makes it so that we can restore functionality to any
application that supports the contextlost/contextrestored events. This
will allow for a theoretical complete graceful recovery for the user
with minimal disruption.
Differential Revision: https://phabricator.services.mozilla.com/D205608
Automatic update from web-platform-tests
screen-wake-lock: Remove @Honry from suggested_reviewers (#45374)
Wanming has not worked on this feature for quite some time.
--
wpt-commits: 1d8c853bef12004e1ff3bbe6fb0452a37d8f0cce
wpt-pr: 45374
Automatic update from web-platform-tests
peerconnection: Fire data channel events synchronously
When message events are dispatched asynchronously, it is possible to
receive some of those after the state change events are dispatched.
This causes some messages to disappear when a data channel is closing.
Tests are added to verify that all messages sent are received properly
by the remote side as mandated by the specification.
Bug: chromium:40072842
Change-Id: I196cd5b6cae4ea443369b2d9bc98dad474162b23
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5397720
Commit-Queue: Guido Urdaneta <guidou@chromium.org>
Auto-Submit: Florent Castelli <orphis@chromium.org>
Reviewed-by: Guido Urdaneta <guidou@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1279016}
--
wpt-commits: 4c286eb7e69b65dbc3757e282d3d89bf611959e7
wpt-pr: 45357
Automatic update from web-platform-tests
Add tentative test for negative @page margins.
The css-page spec doesn't ban negative @page margins (in fact, it
doesn't mention negative margins at all), so the right thing may be to
support them. It's being discussed here:
https://github.com/w3c/csswg-drafts/issues/8508
Change-Id: I8f72f23e7bf88d3e2084dd49deaed6ae7d7ade57
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5401734
Reviewed-by: Ian Kilpatrick <ikilpatrick@chromium.org>
Commit-Queue: Ian Kilpatrick <ikilpatrick@chromium.org>
Auto-Submit: Morten Stenshorne <mstensho@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1279020}
--
wpt-commits: 384306e397f9f6ccd1162c7de66ee3278ac1ef35
wpt-pr: 45372
Automatic update from web-platform-tests
layout: Allow transforming inline replaced elements
This requires passing through information about whether or not the
element in question is replaced when checking to see if it's
transformable and transitively all functions that make decisions about
containing blocks. A new FragmentFlag is added to help track this -- it
will be set on both the replaced items BoxFragment container as well as
the Fragment for the replaced item itself.
Fixes#31806.
--
wpt-commits: c9e74b1951ab94eecb9b554486c3d30255d340ce
wpt-pr: 45296
Automatic update from web-platform-tests
[css-anchor-position] Support revert-(layer) in @position-try
@position-try rules cascade to a separate origin, which means both
revert and revert-layer in @position-try revert back to the author
origin.
Bug: 40281995
Change-Id: I08dac7005a6e79aa2a08fc9d8f63cb80a3e077a4
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5376497
Commit-Queue: Rune Lillesveen <futhark@chromium.org>
Reviewed-by: Anders Hartvoll Ruud <andruud@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1278839}
--
wpt-commits: f0277f5f9a8c9d61b8e3b3e8a1df7485ff10ca5d
wpt-pr: 45172
Automatic update from web-platform-tests
webnn: Add the lstmCell operator
This CL adds the IDL and mojo definition for lstmCell, implements
`MLGraphBuilder::lstmCell()` and adds validations on services side. The
DirectML backend implementation is shared with lstm.
This CL also adds lstmCell validation tests in WPT and unit tests in
`WebNNGraphImplTest`.
Bug: 40206287
Change-Id: I8c740ffa43f79d4667bddca93a064ab788cf8129
Cq-Include-Trybots: luci.chromium.try:gpu-fyi-try-win11-qualcomm-rel-64
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5347306
Reviewed-by: Austin Sullivan <asully@chromium.org>
Reviewed-by: ningxin hu <ningxin.hu@intel.com>
Reviewed-by: Reilly Grant <reillyg@chromium.org>
Auto-Submit: Shiyi Zou <shiyi.zou@intel.com>
Commit-Queue: Shiyi Zou <shiyi.zou@intel.com>
Cr-Commit-Position: refs/heads/main@{#1278827}
--
wpt-commits: b563fb446c00828966ef4bd9cf3fcd45891e6a93
wpt-pr: 44944
Automatic update from web-platform-tests
Add tests for text-wrap: balance and br
Related to https://github.com/w3c/csswg-drafts/issues/9112
--
wpt-commits: dad6f5e038ec13e3506e9a869d759b0138131ee3
wpt-pr: 45369
Automatic update from web-platform-tests
Fenced frames: DisableUntrustedNetwork applies to shared workers
We already get this from the partition nonce work that was done
previously. This CL adds a WPT to confirm that the partition nonce
network disabling also works for shared workers created by fenced
frames.
This CL also adds a test to confirm that a shared worker cannot be used
to postMessage() across a fenced boundary. We already get this behavior
because a fenced frame is treated as being cross-origin, which means the
shared worker will not be shared across fenced frame boundaries, even if
being loaded in a document that is same-origin to the fenced frame.
Change-Id: Ib06a266d85aca3a403221c9eddd839184fb5b6ed
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5394299
Reviewed-by: Andrew Verge <averge@chromium.org>
Commit-Queue: Liam Brady <lbrady@google.com>
Cr-Commit-Position: refs/heads/main@{#1278653}
--
wpt-commits: f94851f9d64723dbe75fa8251cf5a68a321ee7e3
wpt-pr: 45332
Automatic update from web-platform-tests
[Invokers] Clean up invokers tests (#45361)
This change aims to tidy up the invokers tests to reduce the amount of
flakey builds from timeouts due to the amount of tests in each file.
It refactors all of the invokers tests to:
- Use loops for repetitive tests
- Try to consistently test the aliases of auto, explicit empty ("") vs
named.
- Mark any tests using clickOn as timeout long, to avoid flakiness in
CI
- Split the invalid no-ops, to reduce individual test time.
Bug: 40943183
Change-Id: I066b0a6e06143e562a3b0f4c7c1bbaeb309b68f6
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5374369
Reviewed-by: Mason Freed <masonf@chromium.org>
Commit-Queue: Joey Arhar <jarhar@chromium.org>
Reviewed-by: Joey Arhar <jarhar@chromium.org>
Auto-Submit: Keith Cirkel <chromium@keithcirkel.co.uk>
Cr-Commit-Position: refs/heads/main@{#1277818}
Co-authored-by: Keith Cirkel <chromium@keithcirkel.co.uk>
--
wpt-commits: 066ad51422c842bcc8c06c4a713821ebd121f690
wpt-pr: 45361
