NEW308658
[iOS] compositing/geometry/fixed-position-composited-page-scale-smaller-than-viewport.html is a flaky image diff
https://bugs.webkit.org/show_bug.cgi?id=308658
Summary [iOS] compositing/geometry/fixed-position-composited-page-scale-smaller-than-...
Diego De La Toba
Reported 2026-02-25 11:36:00 PST
compositing/geometry/fixed-position-composited-page-scale-smaller-than-viewport.html is a flaky image diff on iOS 26.2 HISTORY: https://results.webkit.org/?suite=layout-tests&test=compositing%2Fgeometry%2Ffixed-position-composited-page-scale-smaller-than-viewport.html IMAGE URL: https://build.webkit.org/results/Apple-iOS-26-Simulator-Release-WK2-Tests/308176%40main%20(1896)/compositing/geometry/fixed-position-composited-page-scale-smaller-than-viewport-diffs.html IMAGES: See attached. REPRODUCITON: I was able to reliably reproduce this failure on iOS 26.2 release ToT with the following: run-webkit-tests --no-build --no-retry --no-show-results --exit-after-n-failures=1 --expect-pass --iterations=1000 --force -f --ios-simulator --release compositing/geometry/fixed-position-composited-page-scale-smaller-than-viewport.html --udid < 3 devices > Note: While the command above reliably reproduces the failure, it does still reproduce less often without running fully parallel. I am going to mark expectations as pass imageOnlyFailure while this pends investigation
Attachments
image diffs (55.81 KB, application/zip)
2026-02-25 11:37 PST, Diego De La Toba
no flags
Radar WebKit Bug Importer
Comment 1 2026-02-25 11:36:07 PST
Diego De La Toba
Comment 2 2026-02-25 11:37:38 PST
Created attachment 478477 [details] image diffs
Diego De La Toba
Comment 3 2026-02-25 11:49:40 PST
EWS
Comment 4 2026-02-25 11:51:42 PST
Test gardening commit 308221@main (afbcd0a20b24): <https://commits.webkit.org/308221@main> Reviewed commits have been landed. Closing PR #59436 and removing active labels.
Diego De La Toba
Comment 5 2026-02-25 12:16:14 PST
Looking at the image expected vs actual I am seeing failures when the text is rendering with subtle subpixel antialiasing differences between runs, causing image comparison failures. However, there are no diffs in what the text is actually testing. Removing the text resolves the issue but this could point to an issue with WebKit's rendering? Regardless putting up a PR with this patch since it seems like a viable option to resolve the flake considering the text is arbitrary. Note: Attempted to fix the antialiasing with -webkit-font-smoothing: none and text-rendering: geometricPrecision but neither resolved the inconsistent text rendering on iOS. Removing the text is the most reliable fix.
Diego De La Toba
Comment 6 2026-02-25 12:24:57 PST
Note You need to log in before you can comment on or make changes to this bug.