[webkit-reviews] review requested: [Bug 65769] New renderer for SVGRectElement : [Attachment 115805] Proposed patch
bugzilla-daemon at webkit.org
bugzilla-daemon at webkit.org
Fri Nov 18 07:25:53 PST 2011
Renata Hodovan <reni at webkit.org> has asked for review:
Bug 65769: New renderer for SVGRectElement
https://bugs.webkit.org/show_bug.cgi?id=65769
Attachment 115805: Proposed patch
https://bugs.webkit.org/attachment.cgi?id=115805&action=review
------- Additional Comments from Renata Hodovan <reni at webkit.org>
This new version contains the following:
* Adapting it to the relevant modifications what are commited from the
previous version (e.g.: the removed toPathData(), etc.)
* Few refactoring, mainly related to the calculating of the boundingboxes.
* Merge inflateWithMarkerBounds() and stroke plumping code snippet into one
common inflateWithStrokeandMarkerBounds() function.
* The previous version contained a solution for avoiding the shadow painting
problem on mac. Since Chromium uses the same painting method like mac we needed
to extend the #if PLATFORM(MAC) condition with (PLATFORM(CHROMIUM) &&
OS(MAC_OS_X)).
* Pixel expecteds of the following tests are updated:
* svg/W3C-SVG-1.1/coords-trans-01-b.svg
* svg/custom/percentage-rect.svg
Their results were one pixel slimmer than it was expected. The problematic
rects had transformations and its sizes were defined in percentages.
I dumped a mass of datas and got the following: all the bounding boxes and
transformations paramters remained the same like in the original
RenderSVGPath's implementation. Additionally, I tried to use fillPath() in my
patch instead of fillRect() and the tests were succeeded again. That's in
RenderSVGRect::fillShape() :
context->fillRect(m_boundingBox);
was changed to:
Path tmp;
tmp.addRect(m_boundingBox);
context->fillPath(tmp);
I suppose it means that just the difference of precision of the two methods
was resposible for the fail. Any other idea?
More information about the webkit-reviews
mailing list