Bug 88230 – [GTK] new test svg/stroke/non-scaling-stroke-pattern.svg fails

• Component: Tools / Tests
• Status: RESOLVED FIXED

[János Badics]

This new test: svg/stroke/non-scaling-stroke-pattern.svg fails after introduced in r119391 on Qt, GTK and EFL platforms.
See http://build.webkit.sed.hu/results/x86-32%20Linux%20Qt%20Release%20-%20Qt5-WebKit1/r119392%20(8214)/svg/stroke/non-scaling-stroke-pattern-diffs.html

https://bugs.webkit.org/show_bug.cgi?id=87925

[János Badics]

I'm sorry for copying wrong links. Let me correct:

This new test: svg/stroke/non-scaling-stroke-pattern.svg fails after introduced in r119391 on Qt, GTK and EFL platforms.
See http://build.webkit.sed.hu/results/x86-32%20Linux%20Qt%20Release%20-%20Qt5-WebKit1/r119392%20(8214)/svg/stroke/non-scaling-stroke

https://bugs.webkit.org/show_bug.cgi?id=88198

[Philip Rogers]

I'll have a patch up in a few minutes to fix this.

[János Badics]

(In reply to comment #2)
> I'll have a patch up in a few minutes to fix this.

Thank you for the quick reply and for looking into it.

[Philip Rogers]

Created attachment 145588 [details]
Update test expectations after 119392

[Philip Rogers]

Comment on attachment 145588 [details]
Update test expectations after 119392

Wrong patch.

[Philip Rogers]

Created attachment 145590 [details]
Update test expectations after 119392

[WebKit Review Bot]

Comment on attachment 145590 [details]
Update test expectations after 119392

Clearing flags on attachment: 145590

Committed r119402: <http://trac.webkit.org/changeset/119402>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Philippe Normand]

I guess you want to keep this bug open, right?

[Philip Rogers]

(In reply to comment #9)
> I guess you want to keep this bug open, right?

Absolutely, thanks Philippe!

[Philip Rogers]

Created attachment 152860 [details]
Update TestExpectations

[WebKit Review Bot]

Comment on attachment 152860 [details]
Update TestExpectations

Clearing flags on attachment: 152860

Committed r122903: <http://trac.webkit.org/changeset/122903>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Csaba Osztrogonác]

It is still valid, but CQ closed it when landing expectations update.

[Philip Rogers]

(In reply to comment #14)
> It is still valid, but CQ closed it when landing expectations update.

Is it still valid? I thought one of the refactorings I've been doing in RenderSVGShape fixed it, as the test is now passing on Saf/mac, Cr/linux, Cr/win, and the QT/EFL/GTK bots look to be happy.

[Raphael Kubo da Costa (:rakuco)]

(In reply to comment #15)
> (In reply to comment #14)
> > It is still valid, but CQ closed it when landing expectations update.
> 
> Is it still valid? I thought one of the refactorings I've been doing in RenderSVGShape fixed it, as the test is now passing on Saf/mac, Cr/linux, Cr/win, and the QT/EFL/GTK bots look to be happy.

For some reason, the test started being flaky on the EFL bots since r129371 on the WK2 bot: <http://build.webkit.org/builders/EFL%20Linux%2064-bit%20Debug%20WK2/builds/3534/steps/layout-test/logs/stdio> and r129372 on the WK1 bot: <http://build.webkit.org/builders/EFL%20Linux%2064-bit%20Debug/builds/6296/steps/layout-test/logs/stdio>

[Raphael Kubo da Costa (:rakuco)]

(In reply to comment #16)
> (In reply to comment #15)
> > (In reply to comment #14)
> > > It is still valid, but CQ closed it when landing expectations update.
> > 
> > Is it still valid? I thought one of the refactorings I've been doing in RenderSVGShape fixed it, as the test is now passing on Saf/mac, Cr/linux, Cr/win, and the QT/EFL/GTK bots look to be happy.
> 
> For some reason, the test started being flaky on the EFL bots since r129371 on the WK2 bot: <http://build.webkit.org/builders/EFL%20Linux%2064-bit%20Debug%20WK2/builds/3534/steps/layout-test/logs/stdio> and r129372 on the WK1 bot: <http://build.webkit.org/builders/EFL%20Linux%2064-bit%20Debug/builds/6296/steps/layout-test/logs/stdio>

And then the flakiness mysteriously stopped in r129380 for the WK1 bot and 129377	for WK2...

[Philip Rogers]

(In reply to comment #17)
> (In reply to comment #16)
> > (In reply to comment #15)
> > > (In reply to comment #14)
> > > > It is still valid, but CQ closed it when landing expectations update.
> > > 
> > > Is it still valid? I thought one of the refactorings I've been doing in RenderSVGShape fixed it, as the test is now passing on Saf/mac, Cr/linux, Cr/win, and the QT/EFL/GTK bots look to be happy.
> > 
> > For some reason, the test started being flaky on the EFL bots since r129371 on the WK2 bot: <http://build.webkit.org/builders/EFL%20Linux%2064-bit%20Debug%20WK2/builds/3534/steps/layout-test/logs/stdio> and r129372 on the WK1 bot: <http://build.webkit.org/builders/EFL%20Linux%2064-bit%20Debug/builds/6296/steps/layout-test/logs/stdio>
> 
> And then the flakiness mysteriously stopped in r129380 for the WK1 bot and 129377    for WK2...

This may have been some other test leaving bad state around. I watch this code pretty closely and I don't think there have been any changes here.

Mind if we close this again?

[Jussi Kukkonen (jku)]

(In reply to comment #18)
> This may have been some other test leaving bad state around. I watch this code pretty closely and I don't think there have been any changes here.
> 
> Mind if we close this again?

The test is _very_ flaky on all EFL and GTK bots. I think that's perfectly valid reason to keep a bug open and as this bug wasn't used to fix any particular issue so far it could as well be this one, right?

[Jussi Kukkonen (jku)]

I can reproduce this with --iterations=3 (or larger). For some reason the third one and all subsequent iterations will always fail here.

The failure is interesting: there are visual artifacts around the square (I assume this is the pattern area). These artifacts change depending on the number of iterations I tell test runner to do, but they are totally reproducible: e.g. the result for iteration 4 will always be the same.

[Jussi Kukkonen (jku)]

(In reply to comment #20)
> The failure is interesting: there are visual artifacts around the square (I assume this is the pattern area). These artifacts change depending on the number of iterations I tell test runner to do, but they are totally reproducible: e.g. the result for iteration 4 will always be the same.

This works in minibrowser as well: refresh the window a few times and the problems start appearing.

[Diego Pino]

The test(s) filed under this bug have been consistently passing for the last 4000 revisions. Marking bug as fixed.

Committed r263254: <https://trac.webkit.org/changeset/263254>

[Radar WebKit Bug Importer]

<rdar://problem/64522650>