http://build.webkit.org/results/Apple%20Mavericks%20Release%20WK1%20(Tests)/r167820%20(5521)/results.html http://webkit-test-results.appspot.com/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=editing%2Finserting%2Ftyping-space-to-trigger-smart-link.html
<rdar://problem/16730393>
-PASS: the anchor for 'www.foo.com' has been created. +Failed: the expected content was 'The <a href="http://www.foo.com">www.foo.com</a> should be underlined and there is an anchor node created for it.', but the actual result was 'The www.foo.com should be underlined and there is an anchor node created for it.'.
Marked in http://trac.webkit.org/changeset/167827 for WK1
Assuming the test is failing because we're hitting the reentrancy guard added by r167818, we should probably break runTest() in the layout test into multiple pieces (using setTimeout() calls). Yep, that fixes the test.
(In reply to comment #4) > Assuming the test is failing because we're hitting the reentrancy guard added by r167818, we should probably break runTest() in the layout test into multiple pieces (using setTimeout() calls). > > Yep, that fixes the test. Correction, I must not have been running against the build of WebKit I thought I was.
Created attachment 230261 [details] Incorrect fix: use setTimeout() in test Perhaps the re-entrant behavior needs to be fixed in WebCore itself.
Created attachment 231199 [details] Patch
Comment on attachment 231199 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=231199&action=review > Source/WebCore/editing/ApplyStyleCommand.cpp:816 > for (size_t i = 0; i < runs.size(); i++) { > - removeConflictingInlineStyleFromRun(style, runs[i].start, runs[i].end, runs[i].pastEndNode); > - runs[i].positionForStyleComputation = positionToComputeInlineStyleChange(runs[i].start, runs[i].dummyElement); > + InlineRunToApplyStyle& run = runs[i]; The better way to write this is: for (auto& run : runs) {
Committed r168641: <http://trac.webkit.org/changeset/168641>