Per the flakiness dashboard, this has been failing reliably on Linux (debug) since r126717: <http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=inspector%2Fstyles%2Fmedia-queries.html> It also became flaky on Linux, Linux 32 and Win (dbg) around that revision.
It isn't chromium specific bug at all, this test started to fail on Qt 32 bit release too: --- /ramdisk/qt-linux-release/build/layout-test-results/inspector/styles/media-queries-expected.txt +++ /ramdisk/qt-linux-release/build/layout-test-results/inspector/styles/media-queries-actual.txt @@ -14,8 +14,8 @@ [expanded] media="all" (media-queries.html) @import screen (media-queries.css:3) -@media not aural (media-queries-1.css:5) -#main { (media-queries-1.css:6) +@media not aural (resources/media-queries-1.css:5) +#main { (resources/media-queries-1.css:6) border: 1px solid black; [expanded] It's strange, but it passes in debug mode and on 64 bit platforms.
Skipped on Qt by https://trac.webkit.org/changeset/126753 Please unskip it with the proper fix.
Antti, Kling? Could you check this regression you caused, please?
(In reply to comment #3) > Antti, Kling? Could you check this regression you caused, please? ping again ...
ping again and again ... Are you going to fix the regression you caused?
This seems like some timing related bug in the inspector or inspector test harness. I couldn't really figure out what that test is doing. Maybe apavlov has ideas what this could be about?
(In reply to comment #6) > This seems like some timing related bug in the inspector or inspector test harness. I couldn't really figure out what that test is doing. Maybe apavlov has ideas what this could be about? As to the test, it makes sure the media query "stack" is dumped for each matched CSS rule (if applicable). Now, I have little knowledge about this particular type of flakiness, but we've seen it occasionally for multiple types of tests that involved building/dumping links for resource URLs. IIRC, some time ago it was a result of the main resource loading not finished by the moment a URL linkification took place. Not sure about the current reason though, as much have changed in this area since then (I remember at least one attempt to fix it by a change to the test harness).
> It's strange, but it passes in debug mode and on 64 bit platforms. This happens on Mac (debug, 64-bit), too: http://build.webkit.org/results/Apple%20Lion%20Debug%20WK2%20(Tests)/r128507%20(3413)/inspector/styles/media-queries-pretty-diff.html Should we roll out r126717 to confirm that it's indeed the revision causing this?
Created attachment 164187 [details] rollout Since it's been several weeks without much progress, I'm going to roll out once EWS reports green. There was one other patch that depended on this change, and it also gets rolled out.
Attachment 164187 [details] did not pass style-queue: Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'Source/WebCore/ChangeLog', u'Source/WebCor..." exit_code: 1 Source/WebCore/css/StyleRuleImport.cpp:108: Should have only a single space after a punctuation in a comment. [whitespace/comments] [5] Total errors found: 1 in 11 files If any of these errors are false positives, please file a bug against check-webkit-style.
*** Bug 95066 has been marked as a duplicate of this bug. ***
I could reproduce the flakiness locally when running this test alone, and rollout resolved that. run-webkit-tests inspector/styles/media-queries.html --repeat-each 100 -v
Comment on attachment 164187 [details] rollout Rejecting attachment 164187 [details] from commit-queue. Failed to run "['/mnt/git/webkit-commit-queue/Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '-..." exit_code: 1 ERROR: /mnt/git/webkit-commit-queue/LayoutTests/ChangeLog neither lists a valid reviewer nor contains the string "Unreviewed" or "Rubber stamp" (case insensitive). Full output: http://queues.webkit.org/results/13854528
Committed manually in <http://trac.webkit.org/changeset/128637>.