Comment on attachment 794759 [details] [diff] [review] Part 1: Forward-declare nsIPrincipal I don't understand how this can compile given that we have a nsCOMPtr<nsIPrincipal> in struct SurfaceFromElementResult and some methods here return SurfaceFromElementResult by value. It seems you're just relying on some of the remaining #includes to get nsIPrincipal.h instead (AFAICT: nsIPrincipal.h <- imgRequest.h <- imgRequestProxy.h <- nsStyleStruct.h <- nsRuleNode.h <- nsStyleContext.h <- nsIFrame.h) So I think this dependency should stay explicit.

Comment on attachment 794760 [details] [diff] [review] Part 2: Move some macros from nsHTMLReflowState.h to nsIFrame.h so that we can stop #including nsHTMLReflowState.h in nsLayoutUtils.h r=mats

(In reply to Mats Palmgren (:mats) from comment #3) > Comment on attachment 794759 [details] [diff] [review] > Part 1: Forward-declare nsIPrincipal > > I don't understand how this can compile given that we have a > nsCOMPtr<nsIPrincipal> in struct SurfaceFromElementResult and some methods > here return SurfaceFromElementResult by value. > > It seems you're just relying on some of the remaining #includes to get > nsIPrincipal.h instead (AFAICT: nsIPrincipal.h <- imgRequest.h > <- imgRequestProxy.h <- nsStyleStruct.h <- nsRuleNode.h <- nsStyleContext.h > <- nsIFrame.h) > > So I think this dependency should stay explicit. Hmm, yeah, you're right.