[webkit-reviews] review requested: [Bug 67988] Large canvas fills should not crash or create unnecessarily large image buffers : [Attachment 107593] Patch

[bugzilla-daemon at webkit.org]

Ben Wells <benwells at chromium.org> has asked  for review:
Bug 67988: Large canvas fills should not crash or create unnecessarily large
image buffers
https://bugs.webkit.org/show_bug.cgi?id=67988

Attachment 107593: Patch
https://bugs.webkit.org/attachment.cgi?id=107593&action=review

------- Additional Comments from Ben Wells <benwells at chromium.org>
This patch has added the early exit. This makes the code much nicer and also
makes the diff cleaner.

The test output has been trimmed, and now doesn't need scrolling when showing
up in test results. I've compromised with the test output: passing lines are
all shown, but failure lines have extra text (...EXPECTED x, GOT y versus just
...x) so they stand out. The javascript for the test has also been cleaned up
based on offline feedback to be more readable.

About the normalised rectangles: as long as Path.boundingRect returns a
normalised rect this code should be fine. Looking at the skia implementation
this will be the case; I cannot tell for sure about CG. Is this a safe
assumption to make? From my reading of the SVG path code
(rendering/svg/RenderSVGPath) it also makes this assumption.