Summary: | [ iOS Release arm64 ] css3/filters/effect-contrast-hw.html is a constant image failure | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Bri Harris <bharris9> | ||||
Component: | CSS | Assignee: | Nobody <webkit-unassigned> | ||||
Status: | NEW --- | ||||||
Severity: | Normal | CC: | ap, sabouhallawa, simon.fraser, webkit-bot-watchers-bugzilla, webkit-bug-importer, zalan | ||||
Priority: | P2 | Keywords: | InRadar | ||||
Version: | WebKit Nightly Build | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Attachments: |
|
Description
Bri Harris
2023-01-19 18:12:54 PST
Created attachment 464571 [details]
Image Diffs for Failing Test
I was able to reproduce this failure on my Apple Silicon Mac at iOS 16 Release ToT running the test as follows: Run-Webkit-Tests css3/filters/effect-contrast-hw.html Scrolling backwards in the history, it appears this test has been a constant image failure since we started running testing on iOS 16. This is a curious diff - it's not the colors that are different, but box sizes! Are contrast filters doing the wrong thing at the edges? It's possible the CoreAnimation's contrast filter has slightly different behavior. But I think it's OK to just increase the pixel tolerance here. Hmm, isn't this going to create user visible cracks in page layout when this filter is used? Pull request: https://github.com/WebKit/WebKit/pull/8897 Test gardening commit 259154@main (9770e5c864b4): <https://commits.webkit.org/259154@main> Reviewed commits have been landed. Closing PR #8897 and removing active labels. It appears that this test has been a constant image only failure on iOS 16 simulator running only on an Apple Silicon machine since this test was introduced at : https://commits.webkit.org/254408@main |