Bug 260556

Summary: fast/dom/selectorAPI/caseID.html is a constant text failure (when repro steps are followed).
Product: WebKit Reporter: Ben Schwartz <ben_schwartz>
Component: DOMAssignee: Chris Dumez <cdumez>
Status: RESOLVED FIXED    
Severity: Normal CC: annevk, cdumez, koivisto, webkit-bot-watchers-bugzilla, webkit-bug-importer
Priority: P2 Keywords: InRadar
Version: WebKit Nightly Build   
Hardware: Unspecified   
OS: Unspecified   
See Also: https://bugs.webkit.org/show_bug.cgi?id=254732

Ben Schwartz
Reported 2023-08-22 16:22:45 PDT
fast/dom/selectorAPI/caseID.html This test is a flaky failure on macOS and iOS. HISTORY (link throwing 403 error): https://results.webkit.org/?suite=layout-tests&test=fast%2Fdom%2FSelectorAPI%2FcaseID.html TEXT DIFF: PASS document.querySelector('#lower1').textContent is 'lower 1' -PASS document.querySelector('#LOWER2').textContent is 'lower 2' +FAIL document.querySelector('#LOWER2').textContent should be lower 2. Threw exception TypeError: null is not an object (evaluating 'document.querySelector('#LOWER2').textContent') PASS document.querySelector('#UPPER1').textContent is 'UPPER 1' PASS document.querySelector('#upper2').textContent is 'UPPER 2' PASS document.getElementById('lower1').matches('#lower1') is true -PASS document.getElementById('lower2').matches('#LOWER2') is true +FAIL document.getElementById('lower2').matches('#LOWER2') should be true. Was false. PASS document.getElementById('UPPER1').matches('#UPPER1') is true PASS document.getElementById('UPPER2').matches('#upper2') is true PASS document.getElementById('lower1').webkitMatchesSelector('#lower1') is true -PASS document.getElementById('lower2').webkitMatchesSelector('#LOWER2') is true +FAIL document.getElementById('lower2').webkitMatchesSelector('#LOWER2') should be true. Was false. PASS document.getElementById('UPPER1').webkitMatchesSelector('#UPPER1') is true PASS document.getElementById('UPPER2').webkitMatchesSelector('#upper2') is true PASS successfullyParsed is true DIFF URL: https://build.webkit.org/results/Apple-Ventura-Release-AppleSilicon-WK2-Tests/267146%40main%20(4573)/fast/dom/SelectorAPI/caseID-pretty-diff.html REPRODUCTION: I was unable to reproduce this bug on macOS Ventura 13.5 (22G74/arm64). I will do more research.
Attachments
Radar WebKit Bug Importer
Comment 1 2023-08-22 16:23:05 PDT
Ben Schwartz
Comment 2 2023-08-22 17:29:54 PDT
REPRODUCTION: It appears that the failure of this test depends on these three tests being run before: fast/dom/HTMLMeterElement/meter-clone.html fast/dom/SelectorAPI/attrname-case-sensitive.xhtml fast/dom/SelectorAPI/caseID-strict.html When running these tests in sequence prior to the reported test, I was able to reproduce this bug on macOS Ventura 13.5 (22G74/arm64) with at a constant rate using the following command: run-webkit-tests --verbose --clobber-old-results --debug --child-processes 1 fast/dom/HTMLMeterElement/meter-clone.html fast/dom/SelectorAPI/attrname-case-sensitive.xhtml fast/dom/SelectorAPI/caseID-strict.html fast/dom/SelectorAPI/caseID.html
EWS
Comment 3 2023-08-22 17:48:26 PDT
Test gardening commit 267166@main (f11bf4050d7b): <https://commits.webkit.org/267166@main> Reviewed commits have been landed. Closing PR #16957 and removing active labels.
Anne van Kesteren
Comment 4 2023-08-23 00:45:08 PDT
Antti, you might want to take a look at this.
Chris Dumez
Comment 5 2023-08-23 16:05:51 PDT
This is caused by the SelectorQueryCache. Disabling that cache makes the issue go away. Likely a regression from when Antti made the SelectorQueryCache be shared across documents.
Chris Dumez
Comment 6 2023-08-23 16:07:13 PDT
(In reply to Chris Dumez from comment #5) > This is caused by the SelectorQueryCache. Disabling that cache makes the > issue go away. Likely a regression from when Antti made the > SelectorQueryCache be shared across documents. I was referring to Bug 254732.
Chris Dumez
Comment 7 2023-08-24 08:56:26 PDT
EWS
Comment 8 2023-08-24 11:45:17 PDT
Committed 267237@main (727dbe30fe5f): <https://commits.webkit.org/267237@main> Reviewed commits have been landed. Closing PR #17016 and removing active labels.
Note You need to log in before you can comment on or make changes to this bug.