The following layout test is failing on Chromium Linux (And is slow on other platforms...) platform/chromium/compositing/filters/background-filter-blur-outsets.html See: http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=platform%2Fchromium%2Fcompositing%2Ffilters%2Fbackground-filter-blur-outsets.html It has been doing so since it's introduction in https://bugs.webkit.org/show_bug.cgi?id=80046 [chromium] Background filters for composited layers I'm marking it as failing in test_expectations.
Er, super weird that this test fails and others don't. Go ahead and I'll see if I can find the difference.
This is a flaky test: 13 function setBlur() { 14 var blurNode = window.document.getElementById('blur'); 15 window.internals.setBackgroundBlurOnNode(blurNode, 10); 16 layoutTestController.notifyDone(); 17 } 18 19 if (window.layoutTestController) { 20 if (window.internals) { 21 window.setTimeout(setBlur, 0); 22 layoutTestController.waitUntilDone(); 23 } 24 layoutTestController.dumpAsText(true); 25 } it's setting a timeout before the <body> and relying the <div id="blur" being parsed when the timer runs. this isn't guaranteed - the timer might fire before we parse the rest of the page. we should run the test code in the load event.
Ooh, thanks.
window.onload = setBlur; just ends up timing out, and I'm not sure why. window.onload = function() { setTimeout(setBlur, 0); } works locally. Is that what you were thinking then James?
Created attachment 138205 [details] Patch layoutTestController.display() ftw! Thanks to enne.
Comment on attachment 138205 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=138205&action=review > LayoutTests/platform/chromium/compositing/filters/background-filter-blur-off-axis.html:29 > + layoutTestController.waitUntilDone(); I don't think you need waitUntilDone/notifyDone anymore if you're doing everything in the onload handler.
Created attachment 138454 [details] Patch right-o
Comment on attachment 138454 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=138454&action=review > LayoutTests/platform/chromium/compositing/filters/background-filter-blur-outsets.html:16 > + window.internals.setBackgroundBlurOnNode(blurNode, 5); Is there a reason to change the blur size on this test? Do you also need to add other platform-specific results?
Comment on attachment 138454 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=138454&action=review > LayoutTests/ChangeLog:11 > + Also, smaller blur for faster test bots. The test was pretty slow on the bots, slow enough to cause race conditions. It seems a smaller blur radius doesn't change the effectiveness but should speed it up some.
(In reply to comment #8) > (From update of attachment 138454 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=138454&action=review > > > LayoutTests/platform/chromium/compositing/filters/background-filter-blur-outsets.html:16 > > + window.internals.setBackgroundBlurOnNode(blurNode, 5); > > Is there a reason to change the blur size on this test? Do you also need to add other platform-specific results? Oh, and I spose I should disable it on !LINUX. Thanks.
*** Bug 83886 has been marked as a duplicate of this bug. ***
Created attachment 138462 [details] Patch
Comment on attachment 138462 [details] Patch Ah, changing the blur to make the test faster makes a lot of sense.
Comment on attachment 138462 [details] Patch Clearing flags on attachment: 138462 Committed r114995: <http://trac.webkit.org/changeset/114995>
All reviewed patches have been landed. Closing bug.