Problem/Motivation
In #3350972: [random test failure] Drupal\Tests\layout_builder\FunctionalJavascript\LayoutBuilderUiTest::testReloadWithNoSections() @nod_ introduced a "fix" for problems with the correct opening of an off-canvas dialog which potentially could fix all JS random test failures for tests using the drupal.dialog.off_canvas somehow.
The basic test is:
- Unskip
BlockFormMessagesTest::testValidationMessage(). - Run _only_ now unskipped
BlockFormMessagesTest::testValidationMessage()a lot of times (usually we go for 1500x) as-is. - Run _only_ now unskipped
BlockFormMessagesTest::testValidationMessage()a lot of times (usually we go for 1500x) _without_ the changes in #3350972: [random test failure] Drupal\Tests\layout_builder\FunctionalJavascript\LayoutBuilderUiTest::testReloadWithNoSections(). - If 2., passes and 3. doesn't, we can safely turn the test back on again.
Per @xjm in #3353085-4: [meta] Determine impact of [#3350972] fix in off-canvas.js on currently disabled FunctionalJavascript tests we now need to go for a 5000-8000 times run for both patches.
Steps to reproduce
Proposed resolution
Remaining tasks
User interface changes
API changes
Data model changes
Release notes snippet
| Comment | File | Size | Author |
|---|---|---|---|
| #3 | 3353088-3.patch | 958 bytes | spokje |
Comments
Comment #2
spokjeComment #3
spokjeComment #4
spokjeComment #6
spokjeComment #7
smustgrave commentedThe condition mentioned in the issue summary seems to have been met. 1500X with the fix passes, 1500x without the fix fails.
Comment #8
spokjeComment #12
spokjeComment #13
spokjeBoth should_fail and non_fail patch run 5 times. That's 7500 individual tests per patch.
Back to RTBC.
Comment #14
xjmI only see the one 1500x job for each? Adding a few more.
Comment #15
spokjeOne job, ran 5 times per patch: https://www.drupal.org/pift-ci-job/2639217 and https://www.drupal.org/pift-ci-job/2639224 (scroll all the way down for the latter).
Comment #16
xjmFixing attribution.
Comment #18
catchCommitted/pushed to 10.1.x, thanks!
Moving to 10.0.x for backport later.
Comment #21
catchCherry-picked to 10.0.x and 9.5.x, thanks!