nsRenderingContext is now a very thin shim around gfxContext; there's no good reason to allocate it on the heap anymore. It might at this point also make more sense if it lived in layout/base rather than gfx/src.

Honestly, I'd almost rather it go away entirely.

I wouldn't mind seeing it go away entirely, but I am not going to have time to do that patch, whereas I might have time to do this one.

Fixing this makes it easier to fix bug 1088760, since we can then convert the code piecemeal and wrap gfxContexts up in in-the-stack nsRenderingContexts at arbitrary points without worrying about allocation overhead.

Added MOZ_STACK_CLASS locally.

This passes Try BTW. https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=1a8a85fb333b

Comment on attachment 8513898 [details] [diff] [review] patch Review of attachment 8513898 [details] [diff] [review]: ----------------------------------------------------------------- I feel that I'm the wrong reviewer, because I've never dealt with nsRenderingContext at all, so I could be missing basic facts about it. Try :jrmuizel or :mattwoodrow.

Comment on attachment 8513898 [details] [diff] [review] patch Review of attachment 8513898 [details] [diff] [review]: ----------------------------------------------------------------- ::: gfx/src/nsRenderingContext.h @@ +16,5 @@ > class DrawTarget; > } > } > > class nsRenderingContext MOZ_FINAL Would MOZ_STACK_CLASS be appropriate here? ::: layout/printing/nsPrintEngine.cpp @@ +2264,5 @@ > fprintf(fd, "Title: %s\n", docStr.get()); > fprintf(fd, "URL: %s\n", urlStr.get()); > fprintf(fd, "--------------- Frames ----------------\n"); > + //nsRefPtr<gfxContext> renderingContext = > + // mPrt->mPrintDocDC->CreateRenderingContext(); Should these lines go away? @@ +3809,5 @@ > fprintf(fd, "URL: %s\n", aURLStr?aURLStr:""); > fprintf(fd, "--------------- Frames ----------------\n"); > fprintf(fd, "--------------- Frames ----------------\n"); > + //nsRefPtr<gfxContext> renderingContext = > + // aDC->CreateRenderingContext(); Same here

Thanks! (In reply to Jeff Muizelaar [:jrmuizel] from comment #8) > Would MOZ_STACK_CLASS be appropriate here? Indeed, see comment 5. As for the commented out lines, probably, but I don't know and I don't care to investigate much (lots of better things to do).

Comment on attachment 8513898 [details] [diff] [review] patch https://hg.mozilla.org/integration/mozilla-inbound/rev/097629b2eb2a