Bug 224910 – REGRESSION (r276377): [ iOS, EWS ] scrollingcoordinator/ios/fixed-inside-sticky-no-stacking-context-2.html is a constant Image Failure

• Status: RESOLVED
• Resolution: FIXED
• Component: Scrolling

[Robert Jenner]

scrollingcoordinator/ios/fixed-inside-sticky-no-stacking-context-2.html

is a constant Image failure on iOS, and is slowing down EWS queues. 

WEBKIT RESULTS HISTORY:
https://results.webkit.org/?suite=layout-tests&test=scrollingcoordinator%2Fios%2Ffixed-inside-sticky-no-stacking-context-2.html

EWS FAILURE:
https://ews-build.s3-us-west-2.amazonaws.com/iOS-14-Simulator-WK2-Tests-EWS/r426730-11881/results.html

IMAGE DIFF LINK:
https://build.webkit.org/results/Apple-iOS-14-Simulator-Release-WK2-Tests/r276393%20(1090)/scrollingcoordinator/ios/fixed-inside-sticky-no-stacking-context-2-diffs.html

Diff images are also attached to this bug below.

[Robert Jenner]

Created attachment 426764 [details]
Expected, Actual, and Diff images for failing test.

Attaching images to this bug.

[Robert Jenner]

Updated test expectations to Failure here to speed up EWS:
https://trac.webkit.org/changeset/276416/webkit

[Robert Jenner]

Was able to reproduce the failure on iOS14 Simulator Release ToT using the following test:

run-webkit-tests --ios-simulator scrollingcoordinator/ios/fixed-inside-sticky-no-stacking-context-2.html -f

There are a lot of revisions that did not get built, so I'm not able to verify a regression point. However, it appears that changes to Scroll Snap at r276382 is the likely cause of the test's failure:
https://trac.webkit.org/changeset/276382/webkit

[Radar WebKit Bug Importer]

<rdar://problem/76995693>

[Martin Robinson]

I took a look at this and I think the culprit here might be r276377. The test starts passing after reverting that commit, but not after reverting r276382.

[Ryan Haddad]

(In reply to Martin Robinson from comment #5)
> I took a look at this and I think the culprit here might be r276377. The
> test starts passing after reverting that commit, but not after reverting
> r276382.
Thanks for looking! I see the test uses "will-change: transform", so that is likely.

[Tim Nguyen (:ntim)]

Created attachment 427242 [details]
Patch

[Tim Nguyen (:ntim)]

Created attachment 427254 [details]
Patch

[EWS]

Committed r276716 (237120@main): <https://commits.webkit.org/237120@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 427254 [details].