We need to go to the parent block to get the column gap, since the RenderMultiColumnSet doesn't hold that value. :)
Created attachment 189346 [details] Patch
Created attachment 189347 [details] Patch
Comment on attachment 189347 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=189347&action=review > Source/WebCore/rendering/RenderMultiColumnSet.cpp:105 > + return parentBlock->style()->fontDescription().computedPixelSize(); // "1em" is recommended as the normal gap setting. Matches <p> margins. The comment makes me think that computedPixelSize() returns a value equivalent to 1em, which is confusing. > Source/WebCore/rendering/RenderMultiColumnSet.cpp:106 > + return static_cast<int>(parentBlock->style()->columnGap()); I think this should use clampToInteger() like roundedIntPoint() etc do. Weird that it's a float.
(In reply to comment #3) > (From update of attachment 189347 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=189347&action=review > > > Source/WebCore/rendering/RenderMultiColumnSet.cpp:105 > > + return parentBlock->style()->fontDescription().computedPixelSize(); // "1em" is recommended as the normal gap setting. Matches <p> margins. > > The comment makes me think that computedPixelSize() returns a value equivalent to 1em, which is confusing. > > > Source/WebCore/rendering/RenderMultiColumnSet.cpp:106 > > + return static_cast<int>(parentBlock->style()->columnGap()); > > I think this should use clampToInteger() like roundedIntPoint() etc do. Weird that it's a float. Let me try just removing the cast completely. Technically now that subpixel layout is turned on, there is no reason to force the gap to be an integer any longer.
Fixed in r143484.