Mostly updating painting functions. Also adding an assert to ensure views aren't called to paint at sub-pixel coordinates, since currently this is never expected behavior.
Created attachment 135565 [details] Patch
Comment on attachment 135565 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=135565&action=review > Source/WebCore/rendering/RenderView.cpp:221 > + // RenderViews should never be called to paint with an offset not on device pixels. > + ASSERT(LayoutPoint(IntPoint(paintOffset.x(), paintOffset.y())) == paintOffset); I feel this will be used at several places to ensure that we properly converted / rounded. Maybe it would be worth having a convenience function to check that (like bool isPixelAlignedLayoutPoint)? > Source/WebCore/rendering/RenderView.cpp:-329 > - // FIXME: Hardcoded offsets here are not good. I think this comment still applies. I read this comment as: the offset is hardcoded to the start of the object padding box (which it still is). I wonder if it really does add something though. > Source/WebCore/rendering/RenderView.h:72 > + virtual void computeRectForRepaint(RenderBoxModelObject* repaintContainer, LayoutRect&, bool fixed = false) const; > + virtual void repaintViewRectangle(const LayoutRect&, bool immediate = false); please decorate the virtual overridden functions as you go with OVERRIDE (not repeated).
Created attachment 136009 [details] Patch for landing
Comment on attachment 136009 [details] Patch for landing Clearing flags on attachment: 136009 Committed r113431: <http://trac.webkit.org/changeset/113431>
All reviewed patches have been landed. Closing bug.
Reopening to attach new patch.
Created attachment 136028 [details] Patch for landing
(In reply to comment #7) > Created an attachment (id=136028) [details] > Patch for landing Removed the offending overrides.
Comment on attachment 136028 [details] Patch for landing Clearing flags on attachment: 136028 Committed r113487: <http://trac.webkit.org/changeset/113487>