Instead of allowing random objects to poke into public member variables on nsXBLPrototypeResources, let's encapsulate them to make code that manipulates them more sane.

We never create any other kind of nsIStyleRuleProcessor than a nsCSSRuleProcessor.

I'm giving this an nsStyleSet-like API, for familiarity.

Comment on attachment 8440479 [details] [diff] [review] Part 1: Give nsXBLPrototypeResources::mRuleProcessor a more concrete type. r=me

Comment on attachment 8440480 [details] [diff] [review] Part 2: Encapsulate nsXBLPrototypeResources::mRuleProcessor. r=me

Comment on attachment 8440481 [details] [diff] [review] Part 3: Encapsulate nsXBLPrototypeResources::mStyleSheetList. >+ binding->PrototypeBinding()->AppendStyleSheetsTo(*array); Why AppendStyleSheetsTo, not just AppendStyleSheets? >+nsXBLPrototypeBinding::RequireResources() Normally we'd call this EnsureResources. r=me with at least the EnsureResources change made.

(In reply to Boris Zbarsky [:bz] from comment #6) > Comment on attachment 8440481 [details] [diff] [review] > Part 3: Encapsulate nsXBLPrototypeResources::mStyleSheetList. > > >+ binding->PrototypeBinding()->AppendStyleSheetsTo(*array); > > Why AppendStyleSheetsTo, not just AppendStyleSheets? I thought AppendStyleSheets would sound like you were passing in the array of sheets that you wanted to append to nsXBLPrototypeResources::mStyleSheetList.

Oh, I see. I totally got what the function did backwards somehow.

https://hg.mozilla.org/integration/mozilla-inbound/rev/d37890e3b9db https://hg.mozilla.org/integration/mozilla-inbound/rev/6038055f6b19 https://hg.mozilla.org/integration/mozilla-inbound/rev/d4e7841a2cac

Missing include: https://hg.mozilla.org/integration/mozilla-inbound/rev/d4edb2cf572f

And a bit more: https://hg.mozilla.org/integration/mozilla-inbound/rev/bff872c9d4b2

https://hg.mozilla.org/mozilla-central/rev/d37890e3b9db https://hg.mozilla.org/mozilla-central/rev/6038055f6b19 https://hg.mozilla.org/mozilla-central/rev/d4e7841a2cac https://hg.mozilla.org/mozilla-central/rev/d4edb2cf572f https://hg.mozilla.org/mozilla-central/rev/bff872c9d4b2

Hey Cameron, this work blocks bug 1025738 which we'd like to backport to Aurora. Is this bug reasonably safe to land there as well? If so, can you please nominate it for approval? Thanks!

Comment on attachment 8440479 [details] [diff] [review] Part 1: Give nsXBLPrototypeResources::mRuleProcessor a more concrete type. Approval Request Comment [Feature/regressing bug #]: - [User impact if declined]: Would prevent bug 1025738 from landing, which solves some memory leak problems. [Describe test coverage new/current, TBPL]: Been on m-c for a couple of weeks. [Risks and why]: Low, this is a small, no-behaviour-change refactoring. [String/UUID change made/needed]: - (request applies to all 3 parts in this bug)

https://hg.mozilla.org/releases/mozilla-aurora/rev/4ee1eccd5cdd https://hg.mozilla.org/releases/mozilla-aurora/rev/df0670536f69 https://hg.mozilla.org/releases/mozilla-aurora/rev/9e809dfdf2ae