1111451 - some codepaths for recreating frames during style change don't start CSS transitions

Status: NEW

(Core :: CSS Parsing and Computation, defect)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

While writing testcases for bug 1110277, I found some deficiencies in the fix for bug 625289.

In particular, there are a number of ways that frames can be recreated during style change processing for which we don't have a ReframingStyleContexts on the stack.  We need to fix that to ensure that:
 (a) we have a ReframingStyleContexts on the stack, and
 (b) we process pending animation restyles after the frame creation.

These codepaths include:

 - the direct call to ProcessRestyledFrames from RestyleTracker::ProcessOneRestyle

 - the calls to RecreatFramesForContent and MaybeRecreateFramesForElement from RestyleManager::RestyleElement.


I think the best solution here involves a bit of refactoring.

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Try runs with the patches I'm about to upload for review:
https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=c1870e1c467a
https://tbpl.mozilla.org/?tree=Try&rev=c1870e1c467a


I'm still hoping to write tests for patches 1, 3, 4, and 5, although:

 - writing tests for patches 1 and 3 requires adding a way to detect whether frame construction has happened; I think this is probably a good idea in general for regression-testing specific performance fixes

 - writing tests for patches 4 and 5 requires finding a specific codepath that I can exercise without triggering bug 1111451, or perhaps fixing bug 1111451 first.


I can't promise to do this immediately, though; I have some other things on my plate as well.  But I'm interested in your feedback on:

 - the above plan

 - whether you think any of the above needs to block landing the patches I'm about to post for review

 - your thoughts on uplifting the patches that are now ready (I'm inclined to want to uplift to both aurora and beta, given the problems that they fix)


Note that if you don't have time to review all the patches, I'd prefer that your prioritize patches 1-3.  I could transfer 4-7 to tn.

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I put comment 1 in the wrong bug; it is now bug 1110277 comment 14.

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

In the revised approach to bug 1110277, I ended up fixing this in patch 6, but I still need to go back and add the tests that I initially didn't add because of this.