Bug 110689 – [GTK] More media tests timing out

• Component: WebKitGTK
• Status: RESOLVED FIXED

[Martin Robinson]

Some more media tests are timing out. This could be related to the new controls styling or to the switch away from WebKit-specific flexible box.

media/controls-after-reload.html
media/controls-strict.html
media/controls-styling-strict.html
media/controls-without-preload.html
media/video-controls-rendering.html
media/video-display-toggle.html
media/video-playing-and-pause.html

[Zan Dobersek]

None of these tests seem to be timing out:
http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20-%20webkit.org&tests=media%2Fcontrols-after-reload.html%20media%2Fcontrols-strict.html%20media%2Fcontrols-styling-strict.html%20media%2Fcontrols-without-preload.html%20media%2Fvideo-controls-rendering.html%20media%2Fvideo-display-toggle.html%20media%2Fvideo-playing-and-pause.html

They started failing with r142947.
http://trac.webkit.org/changeset/142947

I assume these tests should be checked that the controls rendering is as expected and then have proper baselines generated.

[Martin Robinson]

(In reply to comment #1)
> None of these tests seem to be timing out:
> http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20-%20webkit.org&tests=media%2Fcontrols-after-reload.html%20media%2Fcontrols-strict.html%20media%2Fcontrols-styling-strict.html%20media%2Fcontrols-without-preload.html%20media%2Fvideo-controls-rendering.html%20media%2Fvideo-display-toggle.html%20media%2Fvideo-playing-and-pause.html
> 
> They started failing with r142947.
> http://trac.webkit.org/changeset/142947
> 
> I assume these tests should be checked that the controls rendering is as expected and then have proper baselines generated.

I think that they were timing out for me locally because I'm missing some codecs.

[Philippe Normand]

Can this bug be closed then?

[Martin Robinson]

Sure. We need to generate results for them and unskip them though. Before that I guess I should figure out what codecs are missing and ensure they are listed somewhere.

[Zan Dobersek]

Created attachment 190509 [details]
Patch

[Zan Dobersek]

(In reply to comment #5)
> Created an attachment (id=190509) [details]
> Patch

Calvaris, could you check the baselines I've uploaded here? I'd like to land this patch and close the bug, but I'm not really sure whether the controls are rendered as expected.

[Xabier Rodríguez Calvar]

(In reply to comment #6)
> (In reply to comment #5)
> > Created an attachment (id=190509) [details] [details]
> > Patch
> 
> Calvaris, could you check the baselines I've uploaded here? I'd like to land this patch and close the bug, but I'm not really sure whether the controls are rendered as expected.

They look ok at a first glance, but there must be something wrong with your setup, because the fullscreen button of the PNGs you are creating is not the one that I have under my WebKit jhbuild env. Have a look at LayoutTests/platform/gtk/media/controls-styling-strict-expected.png changes.

I suspect that for some reason, you are taking the fullscreen button from somewhere else. I checked with strace and my setup is taking it from the right place and the button is the one in the PNG that is being replaced with your new version.

[Zan Dobersek]

Created attachment 190588 [details]
Patch

[Zan Dobersek]

(In reply to comment #8)
> Created an attachment (id=190588) [details]
> Patch

These baselines seem better, no? There aren't any differences in LayoutTests/platform/gtk/media/controls-styling-strict-expected.png anymore.

[Xabier Rodríguez Calvar]

(In reply to comment #9)
> (In reply to comment #8)
> > Created an attachment (id=190588) [details] [details]
> > Patch
> 
> These baselines seem better, no? There aren't any differences in LayoutTests/platform/gtk/media/controls-styling-strict-expected.png anymore.

Yes, the expectations are ok. What needs to be done now is correcting the changelog, because I can still see a change in controls-styling-strict-expected.png.

[Zan Dobersek]

Ugh, the changelog wasn't updated. Thanks for pointing that out.
Fixed and landed in r144284.
http://trac.webkit.org/changeset/144284

Given that some seem to experience timeouts in these tests, should this bug be left open so someone can investigate that? If not, feel free to close.

[Philippe Normand]

Calvaris had these test timing out too, then somehow solved it by installing some deps listed in the wiki. I'm not sure what was wrong exactly in his setup but I believe the dependencies listed in the wiki are enough to ensure these tests run fine.

[Xabier Rodríguez Calvar]

(In reply to comment #12)
> Calvaris had these test timing out too, then somehow solved it by installing some deps listed in the wiki. I'm not sure what was wrong exactly in his setup but I believe the dependencies listed in the wiki are enough to ensure these tests run fine.

No, it was because updating my system I lost some of the dependencies and some gst plugins were not properly built. After fixing that, they worked properly. I think you can go on.