In https://bugs.webkit.org/show_bug.cgi?id=75552 ('rebaselining some canvas images'), we noted that this test was reporting failure even though the results were "close enough". So I have modified the test to pass, as long as the results are indeed "close enough".
Created attachment 121287 [details]
Tom and Ryosuke- please look at the test change and see if it looks reasonable to you.
If so, Ryosuke, please r+.
LGTM. Allowable difference may grow in the future, but we don't have a quantified range, so let's start conservative.
rniwa has added a few more people in CC, since he isn't very familiar with canvas stuff...
Stephen- could you plese take a look, and r+ if appropriate?
LGTM. I am strongly in favor of this change, as the WinCairo port often has small differences against the Apple Windows port. This would greatly reduce the amount of custom testing needed to stay in sync.
I'm ok with this, as long as it's spec-compliant. r=me
Comment on attachment 121287 [details]
Clearing flags on attachment: 121287
Committed r104293: <http://trac.webkit.org/changeset/104293>
All reviewed patches have been landed. Closing bug.