Currently, the implementation assumes the RenderRegions have a direct node attached to them. This is not true in the case of pseudo-elements renderers.
Created attachment 149947 [details] Patch
Created attachment 161429 [details] Patch
Comment on attachment 161429 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=161429&action=review Just one question about this one. > Source/WebCore/rendering/RenderRegion.h:50 > + // Regions can't have any children. > + virtual bool isChildAllowed(RenderObject*, RenderStyle*) const { return false; } This is weird to me. Eventually we want to turn RenderRegions into RenderBlocks so that they can have ::before and ::after children. In other words, we do want to support children on RenderRegions eventually, so why do we turn that off?
(In reply to comment #3) > (From update of attachment 161429 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=161429&action=review > > Just one question about this one. > > > Source/WebCore/rendering/RenderRegion.h:50 > > + // Regions can't have any children. > > + virtual bool isChildAllowed(RenderObject*, RenderStyle*) const { return false; } > > This is weird to me. Eventually we want to turn RenderRegions into RenderBlocks so that they can have ::before and ::after children. In other words, we do want to support children on RenderRegions eventually, so why do we turn that off? That comment was supposed to be read more like "FIXME PLEASE: Regions can't have any children because they are RenderReplaced and every time in RenderObjectChildList someone tries to add a child on generatedContentContainer an ASSERT will trigger... (make me RenderBlock!)"
Ah ok, got it. Go ahead and put in a FIXME then and put up a new patch and I will r+ and cq+.
(In reply to comment #5) > Ah ok, got it. Go ahead and put in a FIXME then and put up a new patch and I will r+ and cq+. Tomorrow, bed time now.
Created attachment 161683 [details] Patch
Comment on attachment 161683 [details] Patch The patch is a little different. I've realised it's pointless to override isChildAllowed() now and I think this is a better way to do it. Basically if the pseudo-element is a region, the content values are ignored. I think this is inline with the expected behaviour as well. A pseudo-element with flow-from but no content values should still be displayed.
Comment on attachment 161683 [details] Patch r=me
Comment on attachment 161683 [details] Patch Clearing flags on attachment: 161683 Committed r127269: <http://trac.webkit.org/changeset/127269>
All reviewed patches have been landed. Closing bug.
*** Bug 92125 has been marked as a duplicate of this bug. ***
*** Bug 75587 has been marked as a duplicate of this bug. ***