REGRESSION (r288159): Tables layout broken on WPT.FYI

• Component: Tables
• Status: RESOLVED FIXED

[Ahmad Saleem]

Hi Team,

After STP 151 release, I noted that the WPT test case layout is broken on following link:

Link - https://wpt.fyi/results/svg/idlharness.window.html?label=experimental&label=master&aligned&view=subtest

Please refer to attached screenshot for Safari 15.6 behavior vs STP 151 behavior.

Unfortunately, I don't have reduction or which bug caused it or whether it is website not following specs.

Thanks!

[Ahmad Saleem]

Created attachment 461551 [details]
Safari 15.6 vs STP 151

For reference to see broken view.

[Radar WebKit Bug Importer]

<rdar://problem/98537484>

[zalan]

regressed at r288159. preferred width computation.

[Alexey Proskuryakov]

That revision landed 7 months ago, can it be an STP 151 regression?

[zalan]

(In reply to Alexey Proskuryakov from comment #4)
> That revision landed 7 months ago, can it be an STP 151 regression?
certainly not an STP 151 regression.

[Ahmad Saleem]

Can I download STP 150 from any website to check if it is recent regression or it is broken since 7 months?

I really doubt, it is broken since 7 months since even Safari 15.6 includes fixes from Safari TP 148 as per blog post:

Link - https://webkit.org/blog/13009/new-webkit-features-in-safari-15-6/

Which was only released 29 June 2022, approx. two months ago..

[e-mail] - I am not saying you are wrong but I just want to help and give better bug report for it is easier to fix. Hence, just asking this question and stuff. Thanks!

[zalan]

(In reply to Ahmad Saleem from comment #6)
> Can I download STP 150 from any website to check if it is recent regression
> or it is broken since 7 months?
> 
> I really doubt, it is broken since 7 months since even Safari 15.6 includes
> fixes from Safari TP 148 as per blog post:
> 
> Link - https://webkit.org/blog/13009/new-webkit-features-in-safari-15-6/
> 
> Which was only released 29 June 2022, approx. two months ago..
> 
> [e-mail] - I am not saying you are wrong but I just want to help and
> give better bug report for it is easier to fix. Hence, just asking this
> question and stuff. Thanks!
Thank you, I really appreciate it!

[zalan]

Created attachment 461560 [details]
Test reduction

This bugs is specific about about incorrect minimum width computations on content starting with punctuation and no wrapping opportunities (e.g. !INHERITS(getEnclosureList) ) combined with "let's break the content at any arbitrary position (word-break: break-all)".

"x!pass_if_this_line_wraps", is supposed to return the width of 'x!' as the minimum width (can't break in-between "x" and "!". However we completely fail to break the content and return the total content width as the minimum width (so min width == max width). This confuses flexing type of layouts (e.g. table) and produces wide columns.

[zalan]

Created attachment 461571 [details]
Patch

[zalan]

Created attachment 461572 [details]
Patch

[zalan]

Created attachment 461573 [details]
wpt page fixed with the patch

[zalan]

Created attachment 461576 [details]
Patch

[EWS]

Committed 253434@main (17ac2cd774fd): <https://commits.webkit.org/253434@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 461576 [details].