[webkit-reviews] review canceled: [Bug 100026] inconsistency in drawImage with target rect negative dimensions. : [Attachment 171482] Patch

[bugzilla-daemon at webkit.org]

arno. <arno at renevier.net> has canceled arno. <arno at renevier.net>'s request for
review:
Bug 100026: inconsistency in drawImage with target rect negative dimensions.
https://bugs.webkit.org/show_bug.cgi?id=100026

Attachment 171482: Patch
https://bugs.webkit.org/attachment.cgi?id=171482&action=review

------- Additional Comments from arno. <arno at renevier.net>
(In reply to comment #27)
> (In reply to comment #26)
> > (From update of attachment 171482 [details] [details])
> > Attachment 171482 [details] [details] did not pass mac-ews (mac):
> > Output: http://queues.webkit.org/results/14643213
> > 
> > New failing tests:
> > css3/filters/huge-region.html
> > fast/gradients/gradient-on-pseudoelement-crash.html
> > fast/gradients/crash-on-1px-border.html
> 
> Looks like the CSS Filters code and the CSS or Canvas gradients have
problems?

Indeed.
I think this patch should pass the tests. But I'm concerned about this patch:
drawImage/drawImageBuffer is called more than 100 times in many different
times. I investigating them, but in some of the calls, I cannot do much more
than "guess" wether argument can be one day null (sometimes, the check is made
in the caller caller function. Sometimes, I checked that the argument cannot be
logically false). But for example, in the filters, I don't known if 
FilterEffect->asImageBuffer() can under some circumstances be null.