The following layout test is failing on Chromium. media/video-zoom-controls.html Probable cause: http://trac.webkit.org/changeset/94912 http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=media%2Fvideo-zoom-controls.html&showExpectations=true --- e:\b\build\slave\Webkit_Win\build\layout-test-results\media/video-zoom-controls-expected.txt +++ e:\b\build\slave\Webkit_Win\build\layout-test-results\media/video-zoom-controls-actual.txt @@ -20,6 +20,11 @@ RenderDeprecatedFlexibleBox {DIV} at (1,1) size 122x25 RenderBlock {DIV} at (0,0) size 109x43 RenderBlock {DIV} at (109,0) size 13x43 +layer at (245,226) size 0x30 scrollWidth 87 scrollHeight 31 + RenderDeprecatedFlexibleBox {DIV} at (143,9) size 0x30 [color=#FFFFFF] + RenderBlock (anonymous) at (14,0) size 59x31 + RenderText {#text} at (0,1) size 59x29 + text run at (0,1) width 59: "00:00" layer at (113,219) size 13x43 RenderBlock (relative positioned) {DIV} at (0,-9) size 13x43 layer at (246,217) size 51x48 @@ -38,6 +43,11 @@ RenderDeprecatedFlexibleBox {DIV} at (1,1) size 122x25 RenderBlock {DIV} at (0
Asked about this in bug 67861.
Will look. It's surprising to me that a border-image patch would cause this.
I don't fail this test on Lion. If the visual result didn't change maybe just rebaseline? Are you absolutely sure that the border image patch caused this?
(In reply to comment #3) > I don't fail this test on Lion. If the visual result didn't change maybe just rebaseline? > Are you absolutely sure that the border image patch caused this? I think so; I'm using this as my test: ./Tools/Scripts/new-run-webkit-tests -f media/video-zoom-controls.html --print=everything --full-results-html --platform=chromium-gpu-linux -p in a client sync'd to WK @94949 and chromium @100706, and commenting out the line in test_expectations.txt. At those CLs, the command above fails with the text diff shown (and matching the failures in flakiness_dashboard URL included in the OP), and reverting just 94912 (8efc29aa99375a8afb0f1b6e8d711ea09939bf69 in my git checkout) makes the above command-line pass (after rebuilding DumpRenderTree). This seems 100% reproable (i.e. going back and forth between 94912 reverted and not).
Created attachment 116278 [details] Patch
Committed r101027: <http://trac.webkit.org/changeset/101027>