1739391 - [wpt-sync] Sync PR 31504 - Support style invalidation of ':has()' in rightmost compound

Status: RESOLVED FIXED

(Core :: CSS Parsing and Computation, task, P4)

Description

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Sync web-platform-tests PR 31504 into mozilla-central (this bug is closed when the sync is complete).

PR: https://github.com/web-platform-tests/wpt/pull/31504
Details from upstream follow.

Byungwoo Lee <[e-mail]> wrote:

Support style invalidation of ':has()' in rightmost compound

This is the initial CL of supporting style invalidation with ':has()'.

To minimize complexity, we will focus on the following limited usage.

• Supports ':has()' in a rightmost, top-level compound
• Supports ':has()' argument starts with child or descendant combinator
• Only allow attribute or elemental selectors in argument

So the following will be ignored for now.

• Ignore ':has()' in a non-rightmost compound
• Ignore ':has()' in a logical combinations
• Ignore ':has()' argument starts with '~' or '+'
• Ignore all pseudos in ':has()' argument

To support this case, for a change on an element, we need to invalidate
ancestors.

Invalidate every ancestors for every mutations on every element is very
bad for performance.

So we need a way to skip these.

• irrelevant element
• irrelevant ancestor
• irrelevant mutation

To skip the 'irrelevant element' and 'irrelevant ancestor', this CL
introduces two flags in the ComputedStyle.

• AffectingHas : Indicate that a mutation on the element can affect
  ':has()' state.
• AffectedByHas : Indicate that the element can be affected by ':has()'
  state change

During style recalculation, 'AffectedByHas' flag will be set when
SelectorChecker tries to match any ':has()' selector on an element.
For every descendants of the element, the 'AffectedByHas' flag will
be changed to 'AffectingHas' flag and stored.

\<style> .a:has(.b) { color: red }\</style>
\<div>
\<div class="a"> \<!-- AffectedByHas -->
\<div> \<!-- AffectingHas -->
\<div> \<!-- AffectingHas -->
\</div>
\</div>
\</div>
\<div>
\<div>
\</div>
\</div>
\</div>

Only for a mutation on the element with 'AffectingHas', style engine
will walk up ancestors. And it invalidate style only when an ancestor
has 'AffectedByHas'

Still we have invalidation for irrelevant mutation and invalidation on
irrelevant ancestors. We will handle those by separate CLs.

Bug: 669058
Change-Id: Ia7e607e01f9a070c0d69d89679b99efa0a104797
Reviewed-on: https://chromium-review.googlesource.com/3260677
WPT-Export-Revision: e86db0a1e7cb2bec18776d2cbb123cdc2253af88

Comment 1

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Pushed to try (stability) https://treeherder.mozilla.org/#/jobs?repo=try&revision=8f1bb5f1ebb1f2f57cc71954d6b2bc5282ebdc57

Comment 2

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

CI Results

Ran 11 Firefox configurations based on mozilla-central, and Firefox, Chrome, and Safari on GitHub CI

Total 1 tests and 19 subtests

Status Summary

Firefox

OK : 1
PASS: 12
FAIL: 7

Chrome

OK : 1
PASS: 12
FAIL: 7

Safari

OK : 1
PASS: 12
FAIL: 7

Links

Gecko CI (Treeherder)
GitHub PR Head
GitHub PR Base

Details

New Tests That Don't Pass

/css/selectors/invalidation/attribute-or-elemental-selectors-in-has.html
add .child to #div_child: div#div_subject.color: FAIL (Chrome: FAIL, Safari: FAIL)
add .descendant to #div_child: div#div_subject.color: FAIL (Chrome: FAIL, Safari: FAIL)
add .descendant to #div_grandchild: div#div_subject.color: FAIL (Chrome: FAIL, Safari: FAIL)
set descendant to #div_grandchild[attrname]: div#div_subject.color: FAIL (Chrome: FAIL, Safari: FAIL)
change #div_grandchild to #div_descendant: div#div_subject.color: FAIL (Chrome: FAIL, Safari: FAIL)
add descendant to #div_subject: div#div_subject.color: FAIL (Chrome: FAIL, Safari: FAIL)
add "div > descendant" to #div_subject: div#div_subject.color: FAIL (Chrome: FAIL, Safari: FAIL)

Comment 3

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/20439ad7dd9d
[wpt PR 31504] - Support style invalidation of ':has()' in rightmost compound, a=testonly
https://hg.mozilla.org/integration/autoland/rev/532ddd3c5b9c
[wpt PR 31504] - Update wpt metadata, a=testonly

Comment 4

[Marian-Vasile Laza]

https://hg.mozilla.org/mozilla-central/rev/20439ad7dd9d
https://hg.mozilla.org/mozilla-central/rev/532ddd3c5b9c