###!!! ABORT: wrong rule: 'n->GetRule() == aOldAnimRule', file layout/style/nsStyleSet.cpp, line 674

Well, aOldAnimRule is a mozilla::css::AnimValuesStyleRule while n->GetRule() is a mozilla::css::StyleRule... None of the rulenodes up the tree from aOldRuleNode have anything other than a StyleRule hanging off them, in fact. aOldRuleNode seems to be the "visited rulenode" from nsStyleSet::GetContext. In general, it looks like both the animation rule and the mozilla::css::ImportantRule that the non-visited rulenode has are missing from the visited rulenode?

(In reply to Boris Zbarsky [:bz] from comment #2) > In general, it looks like both the animation rule and the > mozilla::css::ImportantRule that the non-visited rulenode has are missing > from the visited rulenode? Tpshe mozilla::css::ImportantRule is supposed to be mising, since it comes from the :link rule in the testcase. But the animation rule being missing is a problem.

And the problem here is that SkipAnimationRules no longer works correctly when !important rules are present (as the FIXME comment in it, in my tree only, suggests). So we get into trouble when the :link style has an !important rule and the :visited one doesn't.

And I have Jesse's test in a separate patch, as a crashtest. I've confirmed that it aborts in the harness without the patch, and works with the patch.

Comment on attachment 806346 [details] [diff] [review] Fix SkipAnimationRule to do the right thing in the presence of new cascading rules. r=me

https://hg.mozilla.org/integration/mozilla-inbound/rev/8af7ea2bad92 https://hg.mozilla.org/integration/mozilla-inbound/rev/b4d1efb548e9

https://hg.mozilla.org/mozilla-central/rev/8af7ea2bad92 https://hg.mozilla.org/mozilla-central/rev/b4d1efb548e9