In CG, drawing an ellipse directly is faster than constructing and filling a rounded rect path. We should optimize fillRoundedRect in the case that it's just drawing an ellipse.
Created attachment 88734 [details] proposed patch
The commit-queue encountered the following flaky tests while processing attachment 88734 [details]: http/tests/inspector/console-websocket-error.html bug 57392 (authors: pfeldman@chromium.org and yutak@chromium.org) The commit-queue is continuing to process your patch.
Comment on attachment 88734 [details] proposed patch Clearing flags on attachment: 88734 Committed r83253: <http://trac.webkit.org/changeset/83253>
All reviewed patches have been landed. Closing bug.
The commit-queue encountered the following flaky tests while processing attachment 88734 [details]: animations/suspend-resume-animation.html bug 48161 (author: cmarrin@apple.com) The commit-queue is continuing to process your patch.
I think this change broke filling circles in canvas when there is a gradient pattern, since the code no longer goes through the fillPath() logic
Why is fillRoundedRect ever drawing an ellipse?
border-radius: 50%
(In reply to comment #8) > border-radius: 50% Is that really the case we're optimizing for? I can see how it would be *much* faster to let CG do the elipse, but seems like an odd motivation for this change.
Drawing circles with css border-radius is fairly common, and some benchmarks to it. I think it's worth the optimization.
OK. I'll have to trust you on this one. :) Does seem like an easy way to draw a circle in CSS.