Summary: | Add more GC tests for ResizeObserver | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Chris Dumez <cdumez> | ||||||||
Component: | Tools / Tests | Assignee: | Chris Dumez <cdumez> | ||||||||
Status: | RESOLVED FIXED | ||||||||||
Severity: | Normal | CC: | rniwa, webkit-bug-importer, ysuzuki | ||||||||
Priority: | P2 | Keywords: | InRadar | ||||||||
Version: | WebKit Nightly Build | ||||||||||
Hardware: | Unspecified | ||||||||||
OS: | Unspecified | ||||||||||
See Also: | https://bugs.webkit.org/show_bug.cgi?id=231235 | ||||||||||
Attachments: |
|
Description
Chris Dumez
2021-10-05 15:16:50 PDT
Created attachment 440276 [details]
Patch
Created attachment 440367 [details]
Patch
Created attachment 440532 [details]
Patch
Can you ensure that GC tests are written in a form like that (https://trac.webkit.org/changeset/261391/webkit)? Testing *one* object's liveness is bad pattern because of conservative GC. Let's test many objects and at least one is dead. (In reply to Yusuke Suzuki from comment #4) > Can you ensure that GC tests are written in a form like that > (https://trac.webkit.org/changeset/261391/webkit)? > > Testing *one* object's liveness is bad pattern because of conservative GC. > Let's test many objects and at least one is dead. This is not a test to make sure that a particular wrapper is GC'd (e.g. a leak test). Therefore, I don't see the issue with using a single wrapper. In other words, if we had a bug, the test would at least flakily fail. If we don't have a bug, the test will consistently pass. Doesn't seem like commit queue is doing anything here?? Comment on attachment 440532 [details]
Patch
Probably because it is the format is too old now. No more ChangeLog.
Pull request: https://github.com/WebKit/WebKit/pull/1682 Committed r295744 (251749@main): <https://commits.webkit.org/251749@main> Reviewed commits have been landed. Closing PR #1682 and removing active labels. |