Bug 259089 – imported/w3c/web-platform-tests/css/css-ui/text-overflow-028.html still fails on iOS after 263428@main

• Status: NEW
• Component: Layout and Rendering

[Karl Rackler]

Created attachment 467016 [details]
text-overflow-028-actual

Description:
imported/w3c/web-platform-tests/css/css-ui/text-overflow-028.html

The test was added at 202007@main, but no way to tell for sure if it has been a consistent failure since then.  This test has been failing since iOS 13, as far back in the dashboard history as 216269@main on 10/10/2019.

This issue can be reproduced using command:
run-webkit-tests --release --iterations=1 --force --ios-simulator  imported/w3c/web-platform-tests/css/css-ui/text-overflow-028.htm

History:
https://results.webkit.org/?suite=layout-tests&test=imported%2Fw3c%2Fweb-platform-tests%2Fcss%2Fcss-ui%2Ftext-overflow-028.html&platform=ios&limit=50000&recent=false

Diff
0.01% image difference.

*Diff images attached*

[Karl Rackler]

Created attachment 467017 [details]
text-overflow-028-diff

[Karl Rackler]

Created attachment 467018 [details]
text-overflow-028-expected

[Radar WebKit Bug Importer]

<rdar://problem/112051313>

[Karl Rackler]

I have marked this test as a consistent image failure while this issue is investigated.

[EWS]

Test gardening commit 265932@main (c2ba4be918b3): <https://commits.webkit.org/265932@main>

Reviewed commits have been landed. Closing PR #15722 and removing active labels.

[Alexey Proskuryakov]

This test used to be marked as failing, but the expectation was removed in 263428@main. 

And it already has fuzziness, attempting to cover for what appears to be an actual bug - we should be able to render the same text twice without differences.

[alan]

> And it already has fuzziness, attempting to cover for what appears to be an
> actual bug - we should be able to render the same text twice without
> differences.
we should, however due to our current run handling/painting, "foobar" does not always pixel-perfect match <span>foo</span><span>bar</span> where individual glyph(s) may slightly move (we have a bug on this somewhere)

("<bdo dir=rtl>ABCDEF</bdo> 123456" vs. "…56 FEDCBA" in text-overflow-28.html)