Bug 254351

Summary: [ iOS, macOS ] imported/w3c/web-platform-tests/css/css-backgrounds/border-image-round-and-stretch.html is a constant ImageOnly Failure.
Product: WebKit Reporter: Bri Harris <bharris9>
Component: Layout and RenderingAssignee: Bri Harris <bharris9>
Status: NEW ---    
Severity: Normal CC: bfulgham, ntim, simon.fraser, webkit-bot-watchers-bugzilla, webkit-bug-importer, zalan
Priority: P2 Keywords: InRadar
Version: Other   
Hardware: Unspecified   
OS: Unspecified   
See Also: https://bugs.webkit.org/show_bug.cgi?id=155955

Comment 1 Bri Harris 2023-03-23 11:18:36 PDT
I was able to replicate this failure on iOS 16 ToT as follows:

run-webkit-tests imported/w3c/web-platform-tests/css/css-backgrounds/border-image-round-and-stretch.html

I was also able to replicate at Ventura ToT the same way.
Comment 2 Radar WebKit Bug Importer 2023-03-23 11:19:06 PDT
<rdar://problem/107144311>
Comment 3 Bri Harris 2023-03-23 14:09:34 PDT
Pull request: https://github.com/WebKit/WebKit/pull/11887
Comment 4 EWS 2023-03-23 14:14:29 PDT
Test gardening commit 262037@main (f2b627192be8): <https://commits.webkit.org/262037@main>

Reviewed commits have been landed. Closing PR #11887 and removing active labels.
Comment 5 Ryan Haddad 2023-03-23 15:53:17 PDT
It looks like this test was re-enabled with https://commits.webkit.org/261903@main.

The fuzzy match metrics added with that commit seem to have made it pass on arm64 based macOS queues, but it continued to fail on x86 macOS bots as well as both arm64 and x86 iOS simulator queues.

The fuzzy match change in https://commits.webkit.org/262037@main may have been too extreme and might cover up real failures.
Comment 6 Brent Fulgham 2023-03-23 15:59:12 PDT
(In reply to Ryan Haddad from comment #5)
> It looks like this test was re-enabled with
> https://commits.webkit.org/261903@main.
> 
> The fuzzy match metrics added with that commit seem to have made it pass on
> arm64 based macOS queues, but it continued to fail on x86 macOS bots as well
> as both arm64 and x86 iOS simulator queues.
> 
> The fuzzy match change in https://commits.webkit.org/262037@main may have
> been too extreme and might cover up real failures.

I think the wider fuzzy-match values are better than the original state of an expected image failure.

It's frustrating that Arm and Intel have different pixel rounding on these images.