So I was debugging the animation failures of bug 1355351, and I thought that we might not be getting a proper animation rule. While looking into it, I found out that we were sending animation-only restyles to the pseudo-element, which is completely bogus because we don't handle those in the traversal (need to send it to the parent instead). Also, I don't see what code would make us follow the post-traversal updating styles in the frames for animation-only restyles, which is the bug bug 1355351 uncovers.
I am looking this.
Apparently we are passing wrong parent element; https://hg.mozilla.org/mozilla-central/file/8e969cc9aff4/servo/components/style/gecko/wrapper.rs#l686
Hi hiro, I'm fixing a bunch of this stuff in bug 1331047, current tip is at https://hg.mozilla.org/try/rev/2f69e6cced796e04c882608def36ce61b198dd49. It needs a bunch more work, but if fixing this is non-trivial, perhaps we can wait for that?
Sure. I will open a new bug for the issue in comment 2. Also, I might be wrong, CSS animations on pseudo element work fine (except for the issue in comment 2). I've been seeing two test cases  failure locally, and I thought it's caused by wrong pseudo element handling, but I noticed it's caused by bug 1336863.  https://hg.mozilla.org/mozilla-central/file/8e969cc9aff4/dom/animation/test/css-animations/file_effect-target.html#l19
Cool, thanks! Yeah, the problem with transitions modulo that is that we actually may miss to update the frame style sometimes (which is what bug 1355351 exposes). I hope to fix that too with bug 1331047, but let's see :)