The following layout test is having an unexpected result on GTK according to the flakiness dashboard:
The interesting bit is that it should not because it's marked as expected [ Pass Failure Timeout ] in WK2 expectations file.
689:webkit.org/b/172452 http/tests/loading/resourceLoadStatistics/grandfathering.html [ Pass Failure Timeout ]
And GTK+ tests are run with those expectations in mind:
> run-webkit-tests --gtk http/tests/loading/resourceLoadStatistics/grandfathering.html
Using port 'gtk-wk2'
Test configuration: <, x86, release>
Placing test results in /webkit/WebKitBuild/Release/layout-test-results
Baseline search path: platform/gtk -> platform/wk2 -> generic
Using Release build
Pixel tests disabled
Regular timeout: 15000, slow test timeout: 75000
Command line: /webkit/Tools/jhbuild/jhbuild-wrapper --gtk run /webkit/WebKitBuild/Release/bin/WebKitTestRunner -
Found 1 test; running 1, skipping 0.
Running 1 test
Running 1 WebKitTestRunner.
The test ran as expected.
Stopping HTTP server ...
Turns out the flakiness dashboard is assuming GTK ports are WK1, therefore WK2 expectations are not taken into account.
Look at the top: Expectations for GTK_LINUX_WK1-RELEASE
This is the result of a logic in determineBuilderPlatform() that determines whether a platform is WK2 or WK1 based on its name: If it does not contain WK2, it assumes it's WK1.
Created attachment 329132 [details]
Comment on attachment 329132 [details]
Clearing flags on attachment: 329132
Committed r225819: <https://trac.webkit.org/changeset/225819>
All reviewed patches have been landed. Closing bug.
The hosted version continues to run an old revision:
Note it still says WK1. Who should I ping for fixing that?
Try asking email@example.com