#1890502: WYSIWYG: Add CKEditor module to core was committed with pretty solid test coverage; one of the things it tests are the available CKEditor plugins out-of-the-box. This uses PluginManagerBase::getDefinitions(), which does not provide any guarantees on the order in which the definitions are returned. For some unknown reason, it used to pass just fine at #1890502: WYSIWYG: Add CKEditor module to core, but now, that's no longer the case. (I don't know what the root cause is, but clearly plugin discovery is either stateful or non-deterministic.)
Patch test failures at:
| Comment | File | Size | Author |
|---|---|---|---|
| #2 | 1912028-1-fail.patch | 1.15 KB | wim leers |
| #2 | 1912028-1-pass.patch | 1.27 KB | wim leers |
Comments
Comment #1
tim.plunkettDupe of #1912026: CKEditorPluginManagerTest fails because plugins are discovered in slightly different order. Double click, I think :)
Comment #2
wim leersThe "pass" patch should pass and provides the fix.
The "fail" patch should fail, yet only contains a new comment.
Comment #3
wim leers#1: ugh! That explains why I got an "issue created" status message upon attaching a file, LOL!
Now, since the patches are here, I'm going to mark the other one as a dupe.
Comment #4
amateescu commentedYes please!
Comment #5
wim leersThe results in #2 for the "fail" patch indicate that
getDefinitions()is non-deterministic. (Or that's at least what it looks like, I don't know anything about the internals of the plugin system, so…).In any case, please commit
1912028-1-pass.patchASAP to unblock lots of inappropriately failed patch reviews in the VDC issue queue and elsewhere :)Comment #6
fabianx commented#2: 1912028-1-fail.patch queued for re-testing.
Comment #7
ParisLiakos commented#1911178: Remove hook_exit() is a victim as well
Comment #8
webchickCommitted and pushed to 8.x. Thanks!
Comment #9
wim leers