Closed Bug 1295258 Opened 6 years ago Closed 6 years ago
Update existing reftests from CSSWG
58 bytes, text/x-review-board-request
This bug was created as a sub part of the fix for Bug 698783 and represents a fix for part 2(a) from David Baron's comment on the original bug. Copying David Baron's comment here for Bug 698783 -> Bug 698783 - Updating the existing tests from CSS Working Group (Step 2a from dbarons comment#12) https://reviewboard.mozilla.org/r/69464/#review66906 So as we discussed earlier today, I think there are a few things to do here that should be separated into a few different patches: (1) we should get rid of the _moz_ in the internal constants for these properties, anytime. That's just code cleanup. (2) we should import the multicol reftests from the CSSWG test repository. This has a few steps: (a) rerun layout/reftests/w3c-css/import-tests.py , and if there are any changes, make a patch with them and push it to try. (b) then, add importing of the multicol tests, and do the same. This depends on (2a). (c) evaluate the failures of the multicol tests added in (b) and figure out if we're doing worse than other browsers that have unprefixed. This depends on (2b). (3) write a patch to layout/style/nsCSSPropList.h to remove the -moz- prefixes there, but add them back as aliases in layout/style/nsCSSPropAliasList.h, and then make the same changes to layout/style/test/property_database.js and any other tests that show failures (probably a small number). This depends on (2c) to decide whether we're ready to do it. (4) Substitute all of the rest of the tree to remove the -moz- prefixes for these properties. This depends on (3). (5) Remove the aliases introduced in (3). This depends on (3) and (4), and shouldn't be done until after we've successfully shipped (3). All of these (1, 2a, 2b, 2c, 3, 4, and 5) could be different bugs, although it's ok if 2a and 2b are in a single bug, and it's also ok if 3 and 4 are in a single bug, or even 1, 3, and 4. But things that happen at different times should be in separate bugs even if they could be in the same bug if done at the same time.
Comment on attachment 8781278 [details] Bug 1295258 - Updating the existing reftests from CSSWG https://reviewboard.mozilla.org/r/71698/#review69240
Attachment #8781278 - Flags: review?(dbaron) → review+
Pushed by firstname.lastname@example.org: https://hg.mozilla.org/integration/autoland/rev/fc8db9256960 Updating the existing reftests from CSSWG r=dbaron
Note that this bug only affects one (previously-imported) reftest, and that reftst is actually disabled right now (with "skip" in the manifest). Hence, there's no strong reason to Try-test this patch before landing, because it doesn't affect tests that we actually run. (One reason this test might be disabled: it references files that we don't seem to have in our tree, since they live in a subdirectory called "iframe" that our python script doesn't seem to pull down. That issue predates this bug here, though. Perhaps our python script should pull that directory down? I filed bug 1295804 on that.)
You need to log in before you can comment on or make changes to this bug.