|Submitter||Diff||Changes||Open Issues||Last Updated|
|Error loading review requests:|
59 bytes, text/x-review-board-request
|Details | Review|
Sorry for yet another reftest-stylo.list change. The reftest expectation changed again. Looks like there are some load failures for no clear reason: https://treeherder.mozilla.org/#/jobs?repo=try&revision=5f49635d40df374dfb7da87dc0268d6539b45a4f I skip them and it's all green except an Rs3 intermittent: https://treeherder.mozilla.org/#/jobs?repo=try&revision=f1b0e92d6b682716e17fbed6c7d4b8bc4bbce9d0 This blocks me from landing Bug 1344132 (again!). I'll land this first and address the unknown failures when I deal with 1351548.
Comment on attachment 8855610 [details] Bug 1354404 - Update reftest-stylo.list expectations. https://reviewboard.mozilla.org/r/127460/#review130156 I guess the problem with these test loads in the fragment is that by using the same test and reference file, the second load ends up just scrolling the test file rather than loading a whole new document, and that's the reason we have some manually specified reference files in the reftest-stylo.list manifests that are different from the test file. In your followup where you will get these tests unskipped again, maybe you could make the reftest harness just not automatically use the test file as the reference file if it finds either the test or reference has a URL fragment? That would make it do a "normal" reftest comparison, rather than Gecko to Stylo, but at least we could still be running these tests. (And the first one you've marked skip specifically helped me in my bug 1345695 investigations, so it would be good to have it re-enabled again soon.