Improve robustness of controls-without-preload by seeking to the start explicitly at the end of the test.
Created attachment 85974 [details] Patch
Comment on attachment 85974 [details] Patch Nice change, thanks!
Committed r81280: <http://trac.webkit.org/changeset/81280>
Odd, this caused a failure on the Leopard release bot: http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r81306%20(29853)/media/controls-without-preload-pretty-diff.html So Sam checked in new results: https://bugs.webkit.org/show_bug.cgi?id=56522
(In reply to comment #4) > Odd, this caused a failure on the Leopard release bot: http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r81306%20(29853)/media/controls-without-preload-pretty-diff.html > > So Sam checked in new results: https://bugs.webkit.org/show_bug.cgi?id=56522 Nah. The test was failing like that for a long time. I thought I broke it too, but I looked back and could see this failing as far as r80697: http://build.webkit.org/results/Leopard%20Intel%20Release%20(Tests)/r80697%20%2829439%29/media/controls-without-preload-pretty-diff.html
And here's Sam's conclusion: https://bugs.webkit.org/show_bug.cgi?id=56532
Indeed, the problem is that the fullscreen button isn't showing (it is 16x16, the mute button is 14x12): - RenderButton {INPUT} at (297,4) size 16x16 - RenderButton {INPUT} at (267,6) size 14x12 + RenderButton {INPUT} at (297,6) size 14x12 but Sam's conclusion in https://bugs.webkit.org/show_bug.cgi?id=56532 is for WebKit2. In any case, the failure didn't start with this change.
(In reply to comment #7) > Indeed, the problem is that the fullscreen button isn't showing (it is 16x16, the mute button is 14x12): > > - RenderButton {INPUT} at (297,4) size 16x16 > - RenderButton {INPUT} at (267,6) size 14x12 > + RenderButton {INPUT} at (297,6) size 14x12 > > but Sam's conclusion in https://bugs.webkit.org/show_bug.cgi?id=56532 is for WebKit2. > > In any case, the failure didn't start with this change. So the MediaPlayerPrivate::supportsFullscreen must be returning false for Leopard on this bot? That seems to be the only code path that could diverge.