Functions were duplicated across test files.

Change-Id: I61d7d03fb57547e8f0b52bc4d27e1efc4504280d
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5869120
Reviewed-by: Anders Hartvoll Ruud <[email protected]>
Commit-Queue: Rune Lillesveen <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1356983}

Test that if the frame is 120x60 and restrictions are max 60x60 then the
result is 60x30 as to maintain the original aspect ratio.

Bug: chromium:363544347, webrtc:366067962
Change-Id: Ic6665871d6d9efde4726dd8a41631eca5ed8a7b8
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5853944
Commit-Queue: Henrik Boström <[email protected]>
Reviewed-by: Harald Alvestrand <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1356990}

The Curve448 algorithms are not implemented in any of the major browsers,
so it's better to define its import/export tests in separate files to
avoid unnecessary noise on the tests for the Curve25519 algorithms.

This CL exposes the WebHID API to dedicated workers behind the blink
runtime feature WebHIDOnDedicatedWorkers.

Intent to prototype: https://groups.google.com/a/chromium.org/g/blink-dev/c/y__BOYfZWzI
Spec PR: WICG/webhid#121
Spec issue: WICG/webhid#120
Demo: https://webhid-worker.glitch.me/

Change-Id: If145f34faf9211a54c8b4a15fbe7e903411b8d1d
Bug: 365932453
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5841991
Reviewed-by: Arthur Sonzogni <[email protected]>
Reviewed-by: Yoav Weiss (@Shopify) <[email protected]>
Reviewed-by: Matt Reynolds <[email protected]>
Commit-Queue: Fr <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357046}

The Ed448 algorithm is not implemented in any of the major browsers,
so it's better to define its sign / verify tests in a separate file
to avoid unnecessary noise on the tests for the Ed25519 algorithm.

When `window.open` is performed with 'noopener' a null handle is returned regardless of whether the navigation succeeded.

Bug: 359245744
Change-Id: I027e218b42936bdad5c0f4272f6a613978dc79de
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5871987
Reviewed-by: Andrew Williams <[email protected]>
Commit-Queue: Kyra Seevers <[email protected]>
Reviewed-by: Kyra Seevers <[email protected]>
Auto-Submit: Janice Liu <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357219}

These two cases expect relative colors to be resolved at parse time when it's possible to do so. However, the rest of the cases in this test have been updated per w3c/csswg-drafts#10328 to expect declared values of relative colors to preserve their original forms.

The cases tested that both rgb and color(srgb both result in color(srgb, but this is not valid any more for specified colors. We already have coverage of the same scenario for computed colors in color-computed-relative-color.html, so these subtests can be removed.

This CL updates the explanation for the edge case in
FederatedAuthRequestImpl::CancelTokenRequest() and adds test which
executes the edge case.

BUG=40940748

Change-Id: I800959c721f13364801110eb20b1e1dff0b45059
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5871992
Commit-Queue: Peter Kotwicz <[email protected]>
Reviewed-by: Christian Biesinger <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357306}

This CL also adds WPT and fixes a mojo comment format issue for
scatterND.

Bug: 363761938
Change-Id: Icaa6ba11943f7e89796f13b70e637c7ec496dd75
Cq-Include-Trybots: luci.chromium.try:win11-blink-rel, mac14.arm64-blink-rel, mac14-blink-rel, mac15.arm64-blink-rel, mac15-blink-rel, linux-blink-rel
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5862351
Reviewed-by: Alex Gough <[email protected]>
Reviewed-by: Weizhong Xia <[email protected]>
Commit-Queue: ningxin hu <[email protected]>
Reviewed-by: Rafael Cintron <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357392}

This option was deprecated and has now been removed from recent versions
of webkitgtk_minibrowser (which will cause the browser to not start if
the option is given.)

Correct a mistake in 62fb506. The leading 0x00 bytes are wrong.

Bug: 367764865
Change-Id: I6bb49467bd273aa25fd9332a9e5e5d868bd86bc5
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5872842
Commit-Queue: Rune Lillesveen <[email protected]>
Reviewed-by: Anders Hartvoll Ruud <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357517}

… navigables (#48260)

Align with spec changes introduced in w3c/webdriver-bidi#770

This change adds support for `position: relative` to table `<caption>`.
In addition to adjusting their position according to inset values, table
captions must also establish containing blocks for descendants that are
absolutely positioned.

Co-authored-by: Oriol Brufau <[email protected]>
Signed-off-by: Martin Robinson <[email protected]>

#47707)

preparing for w3c/webrtc-extensions#214
ServiceWorker tests remain in webrtc-extensions for now

BUG=368070320

Change-Id: I4372915e9d661cd4c9905c020acc96d9f4cab975
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5873851
Reviewed-by: Nina Satragno <[email protected]>
Commit-Queue: Peter Kotwicz <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357604}

The X448 algorithm is not implemented in any of the major browsers, so it's
better to define its deriveBits / deriveKey tests in a separate file to avoid
unnecessary noise on the tests for the X25519 algorithm.

Co-authored-by: wpt-pr-bot <[email protected]>

... with focus on single-party auctions.

Change-Id: Ifb65d0b67ebf4293224bd8a12e9fa724b2320de7
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5866311
Reviewed-by: Russ Hamilton <[email protected]>
Commit-Queue: Maks Orlovich <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357738}

…up (#48265)

Even with no browser, `stop_runner()` should still call `cleanup()` in
case there are other resources to clean up.

Fixes #48263

This CL also implements validation tests in WPT and corresponding
`opSupportLimits`.

webmachinelearning/webnn#375 (comment)

Bug: 40206287
Change-Id: Ic184f2f45a9e201f12fd072675077511bfaec797
Cq-Include-Trybots: luci.chromium.try:win11-blink-rel, mac14.arm64-blink-rel, mac14-blink-rel, mac15.arm64-blink-rel, mac15-blink-rel, linux-blink-rel
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5823074
Commit-Queue: Koji Ishii <[email protected]>
Reviewed-by: Alex Gough <[email protected]>
Reviewed-by: ningxin hu <[email protected]>
Reviewed-by: Koji Ishii <[email protected]>
Auto-Submit: Shiyi Zou <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357776}

…nnels.

Differential Revision: https://phabricator.services.mozilla.com/D220431

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1915143
gecko-commit: 5e3767e9fcca6448d4c1161db2623405369f7f7f
gecko-reviewers: chunmin

… & groupId of initial gUM call.

Differential Revision: https://phabricator.services.mozilla.com/D220945

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1876636
gecko-commit: e85bbde80dd818783d861032877d45b5c59333a4
gecko-reviewers: karlt

…n ComputeBSizeValueAsPercentageBasis().

That is, a definite max block-size and an auto block-size should not compute a
definite percentage basis in the block axis.

Note that we accidentally pass
`testing/web-platform/tests/css/css-sizing/intrinsic-percent-replaced-017.html`.
After applying this patch, the test starts to fail, but it will be fixed in the
next part.

Differential Revision: https://phabricator.services.mozilla.com/D222230

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1918576
gecko-commit: e8d914658c039950185a63ec36d5f270c4986008
gecko-reviewers: dholbert

…e its block size.

In a column-oriented flex container, if a flex item's intrinsic inline size or
its descendants' inline size contributions depend on the item's block size,
`NS_FRAME_DESCENDANT_INTRINSIC_ISIZE_DEPENDS_ON_BSIZE` will be set. In this
case, when determining the flex item's cross-size, we need to mark the flex
item's intrinsic inline size as dirty such that its (auto) inline size is
recomputed when constructing its `ReflowInput`.

`testing/web-platform/tests/css/css-sizing/intrinsic-percent-replaced-017.html`
is the test coverage for this patch.

Differential Revision: https://phabricator.services.mozilla.com/D222231

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1918576
gecko-commit: 51fd81b5ced0f6cec53b3b475fa50a9a48010848
gecko-reviewers: dholbert

…uritypolicyviolation event on javascript URL execution.

Previously, the securitypolicyviolation was created because an inline
event handler was attempted to be executed. The inline event handler's
code didn't matter.

Differential Revision: https://phabricator.services.mozilla.com/D222424

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1919229
gecko-commit: 4349180baf89c9539fb7b7d77ab1be3a4feb2353
gecko-reviewers: tschuster

Differential Revision: https://phabricator.services.mozilla.com/D221655

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1917579
gecko-commit: 8a55e8eb5a286c5961ffdb347bc633fbc0e4964b
gecko-reviewers: chunmin

…nd height of the rectangle are valid.

Per spec of `rect()`, the second (right) and third (bottom) values are
floored by the fourth (left) and second (top) values, respectively. This
is to avoid that they don't cross over the other edge. In our
implementation, we just clamp width and height to make sure they are not
negative. The both ways should be identical.

Note that the spec also mentions another way to handle the similar case
in `inset()`, which would like to apply the algorithm mentioned in
"CSS Backgrounds 3 § 4.5 Overlapping Curves rules" to proportionally reduce
the inset effect to 100%. This may cause a different result from
flooring the values. I suspect this is a spec issue:
w3c/csswg-drafts#10870.

Differential Revision: https://phabricator.services.mozilla.com/D222536

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1916646
gecko-commit: 3c43f092dbbf3d5cab1e39e2b450fee1751407e9
gecko-reviewers: dshin

Differential Revision: https://phabricator.services.mozilla.com/D221050

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1916508
gecko-commit: 12b22d6a5e24b656422b71ffe4f0a8ff8dc54a7b
gecko-reviewers: media-playback-reviewers, alwu

Differential Revision: https://phabricator.services.mozilla.com/D222320

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1918838
gecko-commit: 4b595e1843aa64b05f111f363838f3adea239f3a
gecko-reviewers: dshin

…yle.

Plumb it via the StyleBuilder like we do for the writing_mode for the
same reasons.

Differential Revision: https://phabricator.services.mozilla.com/D222328

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1917703
gecko-commit: 67633c2f0a9c228f5c93b815a753fdc45d3024f0
gecko-reviewers: dshin

…48238)

Most of the implementation is lifted verbatim from an existing
`chromedriver` one, except the nonstandard CDP call in `render_as_pdf()`
is replaced by an equivalent one to WebDriver's "print page" [0].

`shrinkToFit` is actually a WebDriver parameter, not a CDP one, so this
PR also fixes some incorrectly scaled screenshots [1].

[0]: https://www.w3.org/TR/webdriver/#print-page
[1]: https://crbug.com/366293137

…n control's inner frame.

Differential Revision: https://phabricator.services.mozilla.com/D222563

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1919220
gecko-commit: 98ea87fd51e1ccc38c9a41d67c039a71df085cd0
gecko-reviewers: layout-reviewers, jfkthame

…ution for fieldset contents.

Differential Revision: https://phabricator.services.mozilla.com/D222591

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1919220
gecko-commit: b51d80a6a0c4c1064f62a9e39c9752b58aec2131
gecko-reviewers: dshin

Depends on D221033

Differential Revision: https://phabricator.services.mozilla.com/D221034

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1916685
gecko-commit: e1aade50c0eb0a2bf43c77266ebf7cc15afbab77
gecko-reviewers: webdriver-reviewers, whimboo

All browsers except Firefox spuriously fail the precondition in
`onload` [0] because their history includes `about:blank` (initially
loaded by WebDriver [1, 2]), followed by `same-url.html`. Avoid encoding
this implementation detail in the test.

Also, add the current state to each assertion to make failure more
informative.

[0]: https://wpt.fyi/results/html/browsers/browsing-the-web/history-traversal/same-url.html?run_id=5145442707046400&run_id=5078811356168192&run_id=5154571592925184&run_id=6326848455966720
[1]: https://www.w3.org/TR/webdriver/#new-window
[2]: https://github.com/web-platform-tests/wpt/blob/3939cbfd/tools/wptrunner/wptrunner/executors/executorwebdriver.py#L811-L818

This cl updates the buildGraphAndCompute method to support subgraphs
with multiple operators. The output precision for such subgraphs is the
sum of the individual operators’ precisions. Basic tests for both
operators have been added.

Bug: 333756077
Change-Id: I959d307127503e3a81c82aba91351e97d2edd8dd
Cq-Include-Trybots: luci.chromium.try:mac14-blink-rel,mac14.arm64-blink-rel,win11-blink-rel
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5799843
Reviewed-by: ningxin hu <[email protected]>
Reviewed-by: Austin Sullivan <[email protected]>
Reviewed-by: David Baron <[email protected]>
Commit-Queue: Shanxing Mei <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357962}

This CL also adds some WPT conformance tests to verify the
implementation.

webmachinelearning/webnn#375 (comment)

Bug: 40206287
Change-Id: I269b8f07647a3ab40fd896d909c5cc060c0f4386
Cq-Include-Trybots: luci.chromium.try:win11-blink-rel, mac14.arm64-blink-rel, mac14-blink-rel, mac15.arm64-blink-rel, mac15-blink-rel, linux-blink-rel
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5836729
Reviewed-by: ningxin hu <[email protected]>
Auto-Submit: Shiyi Zou <[email protected]>
Reviewed-by: Weizhong Xia <[email protected]>
Commit-Queue: Shiyi Zou <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1357985}

The current way `line-clamp: auto` works is by checking at every line
and after every (line) box whether the box has reached the clamp
point's BFC offset, and place an ellipsis or hide the box depending on
it. For line boxes, this comparison requires computing the BFC offset
at which the line will finish, before it is computed naturally as part
of the inline layout algorithm. And if no box ends right at the clamp
offset, which is almost every time, the layout has to be redone.

An alternative way of doing this work would be by always doing a
relayout: the first pass computes which line number to clamp at, and
the second pass can behave the same as `line-clamp: 3`.

A complication is that with `line-clamp: auto`, unlike with
`line-clamp: 3`, it is possible to break between two lineless boxes.
To account for this, we store both a number of lines until clamp, as
well as a number of remaining boxes between the last line and the
clamp point. We pack the lines and the remaining boxes into a
`LineClampData::UntilClamp` struct.

Since the previous implementation ellipsized lines based on whether
they end at the clamp point, it means that a line could have an
ellipsis even when it was followed by a zero-height block. It also
meant that if a line inside a block with a set `height` might not have
an ellipsis even if it was the last line before clamp, because it did
not end at the clamp point. This patch fixes those bugs, and adds (or
in the case of `line-clamp-auto-032.tentative.html`, fixes) tests for
them.

In cases where the last line before clamp is followed by an abspos,
however, it seems like it would be unexpected for the line to not have
the ellipsis. This remains to be discussed with the CSS Working Group
(see w3c/csswg-drafts#10868), but since
there are previously written tests that rely on this behavior, we have
chosen to special case abspos so such a line is ellipsized.

Bug: 40336192, 367892771
Change-Id: I45881ce76b7b23de731ae69ef5c947957cc7cf93
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5868971
Reviewed-by: Koji Ishii <[email protected]>
Commit-Queue: Andreu Botella <[email protected]>
Reviewed-by: Ian Kilpatrick <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358023}

The Request object should expose "duplex"[1], given our support for
streaming requests. Since it doesn't, requests with a body currently
cannot be used to construct other requests.

This CL fixes that by exposing a fixed-value duplex of "half" (as
specified).

PSA: https://groups.google.com/a/chromium.org/g/blink-dev/c/HPhATidOhuw

[1] https://fetch.spec.whatwg.org/#dom-request-duplex

Bug: 40237822
Change-Id: I1fedecf912a8f327da6c63bcb0801796e3961640
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5872928
Commit-Queue: Yoav Weiss (@Shopify) <[email protected]>
Reviewed-by: Kouhei Ueno <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358031}

Co-authored-by: wpt-pr-bot <[email protected]>

* Add Navigation Tracking Mitigations test

* remove file

* Address comments

* clean up

* address comments

* no op

---------

Co-authored-by: Giovanni Ortuño Urquidi <[email protected]>

#48158

Change-Id: Ic3c18dd7ae76077fa2303018918777dc04d82022
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5861351
Commit-Queue: Robert Flack <[email protected]>
Auto-Submit: Mustaq Ahmed <[email protected]>
Reviewed-by: Robert Flack <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358149}

...and also various bid fields that are observable from them at
top-level auction level.

Change-Id: I5ad605a8c9c21563f97dbb8155d9805205f73f80
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5872540
Reviewed-by: Russ Hamilton <[email protected]>
Commit-Queue: Maks Orlovich <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358208}

I didn't put this behind a pref because it was a bit annoying to do with
the parser changes while keeping it fast, and because, at least nesting
wise, this is unlikely to cause compat issues.

There are some complexities for @scope, which I think I got right.

Fixed the tests not to depend on @scope necessarily per the discussion
in web-platform-tests/interop#697.

Differential Revision: https://phabricator.services.mozilla.com/D222817

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1918408
gecko-commit: 0aa1570bfecae9690855f6d6594e79f59593e36c
gecko-reviewers: dshin, webidl, smaug

This reorganizes the CSS cascading for shadow DOM scopes above the
element's scope so that we do everything one scope at a time, rather
than dealing separately with UA shadow pseudo element styles and
::part() styles.  I'm only aware of one case where this changes
currently-exposed behavior (the cascading of :host::part() selectors
relative to other selectors in the same scope), but it also changes the
implementation to be closer to the way the specifications describe
cascading so that it will extend better to additional features in the
future.

This changes the bucketing approach in RuleSet for a shadow
pseudo-element after ::part() by using the bucket for the innermost
pseudo-element rather than the part bucket.  (We should perhaps have
similar changes for ::slotted() as well, but this does not make those.)

These changes are behind the (status:stable) CSSCascadeCorrectScope
flag, in case we need to disable them.  Supporting this flag involves a
bit of code duplication: while MatchHigherScopeRules and
MatchPseudoPartRules probably could have been the same function, the
number and complexity of flag conditions probably would have made them
unreadable, so it seemed better to separate them.

This also fixes (without a flag) an interaction in
::slotted()::picker(select) by checking FollowsSlotted as for other
pseudo-elements.  (::picker(select) has not yet shipped.)  However, this
doesn't actually fix a test of this interaction,
https://crbug.com/366415131 is to follow up on this.

This also (again, without a flag) makes two changes to the bucketing
optimizations in RuleSet for unshipped features:
* bucket ::details-content, which was not previously done
* don't skip bucketing for ::picker(X) where X is not select; instead
  rely on matching to fail later

Fixed: 1479329
Bug: 1469418, 356158098, 366415131
Change-Id: I148f8240dbaa20e33bb455c8b65732881767ca3e
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5851273
Reviewed-by: Daniel Cheng <[email protected]>
Reviewed-by: Rune Lillesveen <[email protected]>
Commit-Queue: Daniel Cheng <[email protected]>
Auto-Submit: David Baron <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358283}

Currently, directory_tests cleanup before hand to be resilient against
other tests not cleaning up properly. But to be more courteous to other
tests, this CL adds cleanup for after the test as well.

Bug: 346991169
Change-Id: Iff4938926e31691aa3417177fe058a64e46eec0d
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5871483
Reviewed-by: Christine Hollingsworth <[email protected]>
Commit-Queue: Nathan Memmott <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358394}

The purpose of this change is start using WebDriver. WebDriver commands
for Compute Pressure are defined in
https://www.w3.org/TR/compute-pressure/#automation

Dedicated workers need to manipulate virtual pressure sources, but
testdriver calls can only be made in the embedded window, so the
existing solution of having '.any.js' files with `META:
global=window,dedicatedworker` does not work anymore. The solution is
to use message passing so that workers make the embedder window invoke
the virtual pressure source calls and use WPT's variants concept to run
the same test under multiple configurations. Due to this limitation
'.any.js' files needs to be converted to '.window.js'. README.md has
been updated to include a detailed guide for creating tests.

WebDriver endpoints and virtual pressure sources work in Window and
Dedicated Worker scopes, but not shared worker ones: we store virtual
pressure source information in WebContentsUserData, which does not
integrate well with shared workers.

Bug: 347031400
Change-Id: I3cb482c0a689e84012735acd6bc9d33981a7d2f9
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5873130
Reviewed-by: Raphael Kubo Da Costa <[email protected]>
Commit-Queue: Raphael Kubo Da Costa <[email protected]>
Reviewed-by: Reilly Grant <[email protected]>
Auto-Submit: Juha J Vainio <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358424}

See whatwg/encoding#336 for details.

Add tree dump tests for ARIA attributes affected by reference
target that can't yet be tested in WPT.

Add WPT validating that ARIA label updates upon DOM mutations and
reference target changes.

Note that aria-owns coverage is omitted since we plan to exclude
it from reference target.

AX-Relnotes: n/a.
Bug: 346835896,353748711
Change-Id: I89f9e0c81d808ac9f1d20035660963d1d09a00aa
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5739315
Commit-Queue: Dan Clark <[email protected]>
Reviewed-by: Joey Arhar <[email protected]>
Reviewed-by: Aaron Leventhal <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358498}

Original design is to serialize all operand to TFLite tensor in a
loop before serializing operations, but this approach is not
suitable for float16 model because some operations (concat, cast and
reshape) support float16 input data type, other operations don't
support float16, the operation kind isn't known when serializing
operand.

The current design is to serialize the input and output operand of
operation while serializing the operation at the same time. For input
operand, insert a TFLite dequantize operator to convert fp16 to fp32
for graph input, constant and intermediate operand if the current
operation doesn't support float16 inference. For output operand,
insert a TFLite cast operator to convert fp32 to fp16 if the operand is
graph output and the current operation doesn't support float16
inference.

Bug: 339654398
Change-Id: Icd03e7a94874bbb1bf915a7485509badf2e59026
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5807389
Reviewed-by: ningxin hu <[email protected]>
Reviewed-by: Reilly Grant <[email protected]>
Commit-Queue: Junwei Fu <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358484}

Differential Revision: https://phabricator.services.mozilla.com/D222957

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1919658
gecko-commit: 90acfa4e2255174cca231aa5a9cb8349f3fd447d
gecko-reviewers: layout-reviewers, emilio

Differential Revision: https://phabricator.services.mozilla.com/D222958

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1919658
gecko-commit: 94601aa58c87e2b2df729f135fddcdcff7973ac2
gecko-reviewers: layout-reviewers, emilio

With the change in bug 1882670, nsFrameLoader::GetBrowsingContext could
be called when we've destroyed the nsFrameLoader.

Differential Revision: https://phabricator.services.mozilla.com/D222153

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1918505
gecko-commit: 29d6035fe8eee3eb31f4d79f06a1394e1aafe094
gecko-reviewers: emilio

Instead of baking tree-effects into view transition snapshots,
render the snapshot without these effects and add them as CSS to the
::view-transition-group pseudo-element.

This is done by changing the effect order in PaintPropertyTreeBuilder, so that these effects are ancestors of the capture effect. In that way
the effects are not applied inside the capture.

Also had to modify draw_property_utils, to ensure some of the surface
computations don't assume that the view-transition capture would later
be painted into an effect surface.

See design document with detailed conversation and alternatives:
https://docs.google.com/document/d/1jSkIqqlrI4rzZ34cTWC-6TB81cspvHxTcpBD6nDCbE4/edit?tab=t.0#heading=h.tf5nf0fynnc2

Impacted CSS properties:
- opacity
- mask
- clip-path
- filter

The following properties are also affected, will need a follow-up patch to finish this:
- clip
- border-radius (when clipping overflow). This is not entirely set up yet, and will follow up.

Bug: 347947051
Change-Id: I8f24adf0c2be89ff673f111f099eac065fd4bf7f
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5789282
Reviewed-by: Arthur Sonzogni <[email protected]>
Reviewed-by: Khushal Sagar <[email protected]>
Commit-Queue: Noam Rosenthal <[email protected]>
Reviewed-by: Vladimir Levin <[email protected]>
Reviewed-by: Kentaro Hara <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358591}

Simulcast was the last major piece of testing missing. With this CL we
should be good to go for shipping this API in a separate CL.

In addition to the new tests, this CL also tweaks some resolutions used
by other tests to avoid odd resolutions. Some encoder implementations
have resolutions alignments (e.g. 15p would become 16p). While not a
problem on Chrome which uses VP8 + libvpx by default, I predict this
would become a problem when the WPTs run on Safari having H264 by
default which would likely trigger resolution alignments so it's best
to use even resolutions when we have no reason not to.

Bug: chromium:363544347
Change-Id: Id1dba2968f3a0ad3e3206d15b117fdd18643ff41
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5878075
Reviewed-by: Harald Alvestrand <[email protected]>
Commit-Queue: Henrik Boström <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358709}

Change-Id: I1ce0317df871090b28230eb8f83eaf8306210c4a
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5852603
Reviewed-by: Anders Hartvoll Ruud <[email protected]>
Reviewed-by: Steinar H Gunderson <[email protected]>
Commit-Queue: Steinar H Gunderson <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358725}

…tylesheets

The initial Firefox support for network events for cached stylesheets fails in 2 known cases:
- when the stylesheet is loaded via @import (no event emitted)
- when the stylesheet is duplicated in several link tags (too many events emitted)

New tests are added to cover those use cases, which should all be fixed by Bug 1916960 on Firefox side.
All the added tests already pass on Chrome canary.

Differential Revision: https://phabricator.services.mozilla.com/D221685

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1917805
gecko-commit: 26bbaffc60227b5aa78392429c957e8598aaf582
gecko-reviewers: webdriver-reviewers, arai, whimboo

…esheet tests

… Korean.

Differential Revision: https://phabricator.services.mozilla.com/D222809

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1919812
gecko-commit: 3acdba1ef9f0314ae7a2f52985dd2f6c61a5bab8
gecko-reviewers: layout-reviewers, emilio

This introduces tree-scoped keywords (not custom idents) for the first
time. To not waste memory, we keep CSSIdentifierValue as a tree-scope-
agnostic value, and instead use the new CSSScopedKeywordValue whenever
a tree-scoped keyword is needed.

Fixed: 368034380
Change-Id: Ib64d1acaf660e2ce6a1357be3cbe8a62e3655978
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5872877
Reviewed-by: Rune Lillesveen <[email protected]>
Commit-Queue: Anders Hartvoll Ruud <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358734}

This is unfortunately incorrect, it means that:

> AudioDecoder.isConfigSupport({codec: "flac", numberOfChannels: 1, sampleRate: 48000});

resolves to a promise that says that it isn't supported (and it should say yes).
Unfortunately, audioDecoder-codec-specific.https.any.js lacked tests for those
two codecs, and the tests added are incorrect. The spec for this is:
https://w3c.github.io/webcodecs/#check-configuration-support, and the check is
intentionally not very thorough: this method is meant to know if this
implementation generally supports a codec, not to check if configure(...) will
success with very specific parameters. In general, isConfigSupported is done
ahead of downloading any media content, so we don't have a description here.

Differential Revision: https://phabricator.services.mozilla.com/D221510

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1917629
gecko-commit: 6d6a00c29a87857dee38383c3ea44df6840de1ba
gecko-reviewers: chunmin

Configure need to error out if there's no description, but isConfigSupported
should say true.

Differential Revision: https://phabricator.services.mozilla.com/D221656

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1917629
gecko-commit: a6af9793cee645db3af42b0078b73082b64cac17
gecko-reviewers: chunmin

Differential Revision: https://phabricator.services.mozilla.com/D23972

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1522790
gecko-commit: ed9818780499bbba8c5b606bd2632985ee3e5991
gecko-reviewers: jgraham

Depends on D23972

Differential Revision: https://phabricator.services.mozilla.com/D204234

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1522790
gecko-commit: d099e86e9a3f35890fbec91fef48fc0115f03480
gecko-reviewers: whimboo

…arguments.

Depends on D206622

Differential Revision: https://phabricator.services.mozilla.com/D206623

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1522790
gecko-commit: 91d70d27459b254b2b9d8cf4b610c09e593ef369
gecko-reviewers: jgraham

…out of bounds" errors.

Differential Revision: https://phabricator.services.mozilla.com/D220937

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1904665
gecko-commit: ec79322c670d30da20483ab27eea6820a944ec1b
gecko-reviewers: webdriver-reviewers, jdescottes

…ilures related to ValueError: Selector. CLOSED TREE

Backed out changeset d0e02650c62f (bug 1915798)
Backed out changeset f971ce67915c (bug 1915798)
Backed out changeset d20f091c1aab (bug 1904665)
Backed out changeset ec79322c670d (bug 1904665)
Backed out changeset 046e38921371 (bug 1904665)

gecko-commit: c6d992bc9d9c44e1933570703c8ca8f0b7ef39a8

…out of bounds" errors.

Differential Revision: https://phabricator.services.mozilla.com/D220937

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1904665
gecko-commit: 56bcb0cc0d7f3bc2fd59229da70adff02fa403f8
gecko-reviewers: webdriver-reviewers, jdescottes

#47914)

* Test reference target usage via the options argument of attachShadow()

* Code review

* Lint

* Code review

* lint

* Uncomment tests

Change-Id: I0e58a853ae14c684083709a0207cce2e4b75fc03
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5877773
Commit-Queue: Maks Orlovich <[email protected]>
Reviewed-by: Russ Hamilton <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358814}

The test has been flaking for the auxclick action where the
middle-click drag is probably causing autoscroll in some platforms.

This CL changes the action to use the right mouse button instead, after
suppressing any context menu invocation.

Bug: 40851596
Change-Id: I4e36b3aceb75d5edcbd0c9599ada9f7af9ef7b78
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5873347
Commit-Queue: Mustaq Ahmed <[email protected]>
Reviewed-by: Kevin Ellis <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1358971}

* Speculativly test :has-slotted(), see: w3c/csswg-drafts#10586

* Speculatively test :has-slotted() flattened through multiple slot elements

* Lint

…te expectation for Android.

Differential Revision: https://phabricator.services.mozilla.com/D223085

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1920301
gecko-commit: 84ea7ef6f9b052ffa607bac7a02f6d709eba8972
gecko-reviewers: jgraham

1. WPT tests now accept `anchor-size()` property for sizing, inset, and
   margin properties [1].
2. Adjust test expectations for `anchor-size()` fallbacks containing
   `anchor-size()` - This will need to be followed up once [2] is resolved.
3. Adjust expectations for devtools' auto-completion related mochitests.
4. We don't yet handle `anchor-size()` in `calc()`.

[1] w3c/csswg-drafts#9827
[2] w3c/csswg-drafts#10831

Differential Revision: https://phabricator.services.mozilla.com/D222535

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1900232
gecko-commit: a0bc0fe119dd86f4c51d31d2aadfa20133e0f595
gecko-reviewers: firefox-style-system-reviewers, devtools-reviewers, nchevobbe, emilio

This PR implements new tests for the import operation of the
ECDH and ECDSA algorithms, defining failure test cases.

Additionally, it applies some refactoring to the code
already used for OKP keys.

The IteratorRecord#return() function exists as an optional method that
sync and async iterator records can supply [1] [2]. They allow for the
language, or any consumer of an iterable, to signal to the iterable that
the consumer will stop consuming values prematurely (i.e., before
exhaustion).

This method must be invoked when the consumer aborts its subscription
to an Observable that was derived from an iterable. The abort reason is
supplied to the `return()` iterator function for completeness. This CL:
  1. Adds tests for sync & async iterables
  2. Implements this for async iterables

A follow-up CL will implement this for sync iterables.

The semantics are specified in
WICG/observable#160.

[1]:
https://tc39.es/ecma262/#table-iterator-interface-optional-properties
[2]: https://tc39.es/ecma262/#table-async-iterator-optional

Bug: 40282760
Change-Id: Ie1091b24b233afecdec572feadc129bcc8a2d4d3
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5854985
Reviewed-by: Mason Freed <[email protected]>
Commit-Queue: Dominic Farolino <[email protected]>
Reviewed-by: Nate Chapin <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1359083}

This is a set of refactors that:
 1. re-points existing WPTs and web_tests from <selectlist> to
    <select style="appearance:base".
 2. renames tests from `-selectlist-` to `-select-`

These are done in batches, to make the review task easier. One CL
will re-point tests (without renaming them) and the next will do
the rename.

These are all in service of removing the <selectlist> prototype.

Bug: 335458492
Change-Id: Ie8dfdc37f1c3a6323d526ca1979d315c111d93f5
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5857249
Reviewed-by: Joey Arhar <[email protected]>
Commit-Queue: Mason Freed <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1359127}

This is a set of refactors that:
 1. re-points existing WPTs and web_tests from <selectlist> to
    <select style="appearance:base".
 2. renames tests from `-selectlist-` to `-select-`

These are done in batches, to make the review task easier. One CL
will re-point tests (without renaming them) and the next will do
the rename.

These are all in service of removing the <selectlist> prototype.

Bug: 335458492
Change-Id: Iac8e531e0c3354e2637350218e6e4e43002f6747
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5857251
Commit-Queue: Mason Freed <[email protected]>
Reviewed-by: Joey Arhar <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1359131}

This CL implements cumulativeSum for DirectML backend.

This CL also adds some WPT conformance tests [1] to verify the
implementation.

[1]:
    https://learn.microsoft.com/en-us/windows/win32/api/directml/ns-directml-dml_cumulative_summation_operator_desc#examples

Bug: 40206287
Change-Id: Id49674a854ee37240a593a4882893e75b0950a8f
Cq-Include-Trybots: luci.chromium.try:win11-blink-rel, mac14.arm64-blink-rel, mac14-blink-rel, mac15.arm64-blink-rel, mac15-blink-rel, linux-blink-rel
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5850648
Commit-Queue: Bin Miao <[email protected]>
Reviewed-by: ningxin hu <[email protected]>
Reviewed-by: Weizhong Xia <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1359137}

Differential Revision: https://phabricator.services.mozilla.com/D223140

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1920101
gecko-commit: 956e3824f9bc74fa76077d01fd870eb275f1ea13
gecko-reviewers: edgul

Differential Revision: https://phabricator.services.mozilla.com/D222963

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=1920104
gecko-commit: df1fc10f7b4d25524907e6682d39b1e0a7b7c8d5
gecko-reviewers: edgul, smaug

…ctor

When encountering a rule with an invalid selector, we currently recover
by consuming until the next semicolon, which isn't correct. Instead,
we should simply do nothing, and keep parsing right after the {}-block.

The spec used to be less clear about this, but at least at the time
of writing, nothing invokes the skip-until-semicolon behavior except
for rule preludes that look like custom properties.

This CL fixes this issue by adding a `invalid_rule_error` out-param
to ConsumeStyleRule. There are two other things that are worth noting:

 - The "consume a qualified rule" algorithm in the spec mentions
   a `nested` flag which modifies its behavior. This flag is
   actually equivalent to `semicolon_aborts_nested_selector`
   in our parser, so I've renamed this.
 - During handling of `custom_property_ambiguity`, nested=false case,
   we're supposed to consume a block, and do nothing.
   This can't be tested using CSSParserImpl's public API, since
   failing to consume a block there would just cause the next attempt
   at parsing a rule to consume an empty prelude (then the block).
   But I don't think we should rely on that, hence I'm using unit
   tests to cover that line.

[1] https://drafts.csswg.org/css-syntax/#consume-qualified-rule

Test: external/wpt/css/css-syntax/custom-property-rule-ambiguity.html
Change-Id: I04dc0c3745be5eafe792fd6c40971545274d9794
Fixed: 343979378
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5881815
Commit-Queue: Anders Hartvoll Ruud <[email protected]>
Reviewed-by: Steinar H Gunderson <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1359288}

Now that directory tests have a cleanup function builtin to clear up any
files left behind, remove all the previous cleanup code that added this.

Mostly a mechanical change of removing the test argument from these
functions:
- createEmptyFile
- createDirectory
- createFileWithContents
- create_file_system_handles

Verified that all these functions appear in directory_test.

Bug: 346991169
Change-Id: I751cd6c3eb5292ac791de872a89abe195490c7e7
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5878964
Reviewed-by: Christine Hollingsworth <[email protected]>
Commit-Queue: Nathan Memmott <[email protected]>
Cr-Commit-Position: refs/heads/main@{#1359545}