This patch adds the ability to parse most CSS filters without a context.
OffscreenCanvas can use this on worker threads to provide support for
filter operations.
Differential Revision: https://phabricator.services.mozilla.com/D179994
The existing function doesn't not work for cases like when mousedown
and mouseup were happened in the same hierarchy but the mousedown target
was a different interactive element than the mouseup target. This
could be easily accomplished by setting `pointer-events: none` or
`display: none` to the `active` pseudo-class of the mousedown target.
This patch fixes this by when the logic tries to find the event target
tree of the mousedown target, it continues building up (by recursively looking
up the parents) the tree instead of stopping at the first interactive element.
Differential Revision: https://phabricator.services.mozilla.com/D178844
Better control of the ABI has its own merit, but in that case, we
significantly shrink the amount of dynamic symbol, which in turns
reduces the size of the binary.
As this limits the exported symbol to one, there's no longer a risk to
conflict with libjs, so just remove the symbol versioning.
On linux64 opt build, this saves ~2.5MB on the installer size libxul.so
Differential Revision: https://phabricator.services.mozilla.com/D179806
Automatic update from web-platform-tests
Make `break-spaces-{006,011}.html` more robust
This patch adds ".1" to the width so that the test can check
the `break-spaces` behavior more robustly without worrying
about floating point errors.
Bug: 911284, 1409445
Change-Id: Ia150ccee08ada462cb1f7c5e6d8b758a5ef7b7db
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4553379
Reviewed-by: Dominik Röttsches <drott@chromium.org>
Reviewed-by: Andreu Botella <abotella@igalia.com>
Commit-Queue: Koji Ishii <kojii@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1147779}
--
wpt-commits: 0c1bc9103126e2e8ef72fad05ad43ebad20fc4a5
wpt-pr: 40163
Automatic update from web-platform-tests
Notifications: test that silent defaults to null
For https://github.com/whatwg/notifications/pull/194.
--
wpt-commits: ac2c4415f6ea3f0d08a2d19f0e104ce447054db1
wpt-pr: 40160
Automatic update from web-platform-tests
Make shorthand serializer for "column-rule"
Make function GetShorthandValueForColumnRule to omit initial values
during shorthand serialization
Bug: 1429866
Change-Id: Ief5c04aed4c24c4f8d8412a978755d08d1cd89b3
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4546749
Reviewed-by: Anders Hartvoll Ruud <andruud@chromium.org>
Commit-Queue: Hyomin Kim <hyoputer.kim@samsung.com>
Cr-Commit-Position: refs/heads/main@{#1147743}
--
wpt-commits: eb55eef546c0cf04a8bd2c783ca6f6bcd72dacc3
wpt-pr: 40161
Automatic update from web-platform-tests
Fix format specifier in webdriver module
--
wpt-commits: e5d3dd0d81879bd700149fc6cd2873a458ed1a56
wpt-pr: 40131
Automatic update from web-platform-tests
typo: navigation -> nav (#40152)
Typo was causing failures due to an element continuation.
--
wpt-commits: 928bfaf1d95c859e8310a014a6486fa8e3961f31
wpt-pr: 40152
Automatic update from web-platform-tests
Add Compression Test to ensure handling large outputs during the flush stage (#40108)
Ensure compression streams handles large flush output during the closing stages
of the compression stream.
--
wpt-commits: 3b32d1c0b5635ccbba172461a68a81d5a85c8614
wpt-pr: 40108
Automatic update from web-platform-tests
[anchor-position] Support anchor-scroll with fragmented containers
We currently check only the first fragment of the containing block for
anchor-scroll, and therefore can fail if the containing block is
fragmented.
This patch fixes it by iterating through all the fragments and check
every NGPhysicalAnchorQuery there. It also provides two utility
functions that are better abstracted, so that clients like
anchor-scroll and possibly DevTools (in the future) don't need to
deal with fragmentation details.
This patch also reveals an issue with relpos inline containing block,
and marks it with TODOs and a currently failing test. It will be fixed
by a followup patch.
Bug: 1446442
Change-Id: Ia08fd094616c80b40d6cbb555ae3153fbfaea04c
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4546546
Reviewed-by: Koji Ishii <kojii@chromium.org>
Commit-Queue: Xiaocheng Hu <xiaochengh@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1147546}
--
wpt-commits: 490713f7fb0da11c5b5da3a65d4573e2cc83e1b4
wpt-pr: 40084
Automatic update from web-platform-tests
Remove FencedFrameConfig URL attribute
We're removing what we colloquially refer to as "default mode" in fenced
frames, which is where you manually construct a `new
FencedFrameConfig()` object with a (usually) non-opaque URL, to use
fenced frames without relying on cross-site data from FLEDGE or Shared
Storage. If a use-case that requires this mode of fenced frames presents
itself in the future, we will re-add it. But for now, this entails:
1. Removing the FencedFrameConfig#url IDL attribute
- Handled by this CL
- See corresponding spec change:
https://github.com/WICG/fenced-frame/pull/88
2. Removing the FencedFrameConfig web-exposed constructor
- This will happen in a subsequent CL
We're removing the URL attribute of the FencedFrameConfig interface [1] because if fenced frames can only be used with FLEDGE/Shared Storage,
the URL member is always "opaque", since the true value depends on
cross-site data. An always "opaque"-returning attribute does not add any
value to web developers.
To carry out this change we must remove tests that rely on observing
the FencedFrameConfig#url attribute, as these tests will need to be
removed anyways when fully removing "default mode".
[1]: https://wicg.github.io/fenced-frame/#fenced-frame-config-interfaceR=gtanzer@chromium.org
Bug: 1123606
Change-Id: Id29e3b9dac9241d84722c3695f59d9865485d4a5
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4547401
Reviewed-by: Nasko Oskov <nasko@chromium.org>
Commit-Queue: Dominic Farolino <dom@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1147330}
--
wpt-commits: 4eab05e5e460d28219d324c13e5adcc94c5db109
wpt-pr: 40093
Automatic update from web-platform-tests
Shared Storage: Add sharedStorageWritable option to fetch
We add the option to mark a request as eligible for writing to Shared
Storage, i.e. we add `sharedStorageWritable` to the `RequestInit` IDL
dictionary for `fetch()` and plumb it through to
`network::ResourceRequest`, subject to a `PermissionsPolicy` that may
change the value of the bit for redirects.
Integration tests will be added in the follow-up CL https://crrev.com/c/4519162.
This is part of a project to add support to writing to shared storage
from response headers. See
https://github.com/WICG/shared-storage/pull/70 for more information
and https://crrev.com/c/4444068 for a working prototype for `fetch()`.
Bug: 1434529,1218540
Change-Id: Ib133492bd90e8cc37b9bc0c7bc61861e0b2afc63
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/4517382
Reviewed-by: Yoav Weiss <yoavweiss@chromium.org>
Commit-Queue: Cammie Smith Barnes <cammie@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1147313}
--
wpt-commits: cd5b4ce47a0e019bb9e5e320a43f7c8e574c007f
wpt-pr: 40037
Automatic update from web-platform-tests
[wptrunner] Wait for TestRunnerManager threads to exit on interrupt (#40086)
When wptrunner is interrupted, the manager group sets its stop flag but
doesn't wait for the runner manager threads to acknowledge and join.
Besides possible resource leakage, this can cause the threads to
continue logging events after the main thread calls `shutdown()` on the
`web-platform-tests` logger, which is not allowed.
In this change, the main thread waits for the threads to join with a
timeout on an interrupt. The timeout needs to be long enough for the
threads to cycle through their event loops one more time, but not so
long as to appear unresponsive.
--
wpt-commits: 4428a332888632faa2f59d0319043e9acbd9afad
wpt-pr: 40086
Automatic update from web-platform-tests
[image-set] Add tests for negative and zero resolutions. (#40129)
calc() with negative resolution is supposed to clamp to zero. Zero
resolutions are supposed to be <invalid-image>'s from the css images
level 4 spec. This adds parsing and rendering test cases for those
situations.
--
wpt-commits: ba004353155163b0c356b753394e50d2dd6f0b87
wpt-pr: 40129
Automatic update from web-platform-tests
Fix networkState_during_loadstart video sub-test.
networkState_during_loadstart video sub-test now uses the video element rather than the audio element from the audio sub-test.
--
wpt-commits: 780ba7d2aef5b5147601b14ced34258ffa53b272
wpt-pr: 40126