Open Bug 769659 Opened 12 years ago Updated 2 years ago

adding some new reftests can break following tests due to "stuck" canvases


(Testing :: Reftest, defect)



(Not tracked)


(Reporter: jfkthame, Unassigned)


I wonder if this is a manifestation of something similar to bug 384631 (or even a dupe), but that's been dormant for long enough that maybe we need to take a fresh look. The logs at

such as for example

show 129 failures on Linux64, but not on other platforms (the individual failures of some font-matching/bold-system-fallback-* tests are expected).

Viewing the log in reftest analyzer, it appears that canvases from the newly-added (and random-on-linux) testcases are being used in place of the proper test or reference in the following tests, leading to massive fail.
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.