Since the last round of repaintRect fixes (SVG now correctly calculates repaint rects), the webkit-shadow CSS stuff mixed with SVG is broken. You have to run "run-webkit-tests --tolerance 0 -p svg" in order to see the differences, as they're below our default tolerance of 1%. I guess the shadowRect calculation in SVGRenderSupport.cpp, prepareToRenderSVGContent() is off. CC'ing Beth as she's the original author.
(In reply to comment #0) > Since the last round of repaintRect fixes (SVG now correctly calculates repaint > rects), the webkit-shadow CSS stuff mixed with SVG is broken. > You have to run "run-webkit-tests --tolerance 0 -p svg" in order to see the > differences, as they're below our default tolerance of 1%. > > I guess the shadowRect calculation in SVGRenderSupport.cpp, > prepareToRenderSVGContent() is off. CC'ing Beth as she's the original author. The latest patches on repaintRect fixes some shadow examples. The remaining failures were introduced with the new Transparencylayer in SVGRenderSupport. I agree Niko, the size of the clipping rect in seems to be wrong. Also the shadow size is not respected, if the opacity is < 1. The clipping rect gets smaller.
Fixed in one of the last revisions.