Introduce Document::body() for call sites interested in the <body> element only (not <frameset> like Document::bodyOrFrameset() does).
Created attachment 245404 [details] Patch
Comment on attachment 245404 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=245404&action=review > Source/WebCore/rendering/RenderLayerBacking.cpp:1793 > + auto* rootObject = renderer().document().documentElement() ? renderer().document().documentElement()->renderer() : nullptr; Surprised you didn’t add a local variable for documentElement here given what you did everywhere else > Source/WebCore/rendering/RenderLayerBacking.cpp:1806 > + auto* bodyRenderer = body ? body->renderer() : nullptr; > + if (!bodyRenderer) > return false; Could just return false one more time a line earlier instead of using ? : here
Comment on attachment 245404 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=245404&action=review >> Source/WebCore/rendering/RenderLayerBacking.cpp:1793 >> + auto* rootObject = renderer().document().documentElement() ? renderer().document().documentElement()->renderer() : nullptr; > > Surprised you didn’t add a local variable for documentElement here given what you did everywhere else documentElement() is a trivial getter. I cached the return value of body() and bodyOrFrameset() because those do tree traversal. >> Source/WebCore/rendering/RenderLayerBacking.cpp:1806 >> return false; > > Could just return false one more time a line earlier instead of using ? : here Ok.
Created attachment 245418 [details] Patch
Created attachment 245419 [details] Patch
Comment on attachment 245419 [details] Patch Clearing flags on attachment: 245419 Committed r179172: <http://trac.webkit.org/changeset/179172>
All reviewed patches have been landed. Closing bug.