In bug 388548 comment 3, bzbarsky wrote: > I really wish we didn't duplicate logic like this between the check > callbacks and the actual computation. :( We can avoid this duplication by initializing |aInherited| better and not setting it in nsRuleNode::Compute*Data in the cases where it's already set based on that initialization. We should probably have some assertions to verify things, though.
Assignee: dbaron → nobody
Status: ASSIGNED → NEW
Target Milestone: mozilla1.9beta1 → Future
Created attachment 273054 [details] [diff] [review] broken patch Here's what I was thinking, roughly. This achieves the assertions by making parentdata_ null when we shouldn't be using it, but there are a bunch of places where we dereference things in parentData because we might use them. So it crashes, probably because of that, although I haven't checked. I'm not going to work further on this now; I can't justify making it a priority.
You need to log in before you can comment on or make changes to this bug.