Our canvas code gives exceptions on r<0 for arc. This behavior matches the spec, but breaks some sites like this above.
Created attachment 28720 [details] silent return for r<0 patch
What’s our strategy on this for canvas? If the spec is wrong I think we need to get the spec fixed too, and not just change our implementation. Have we been doing this same type of thing for other canvas issues?
It's ok to return if r<0, but there is no need to throw an exception here. The same for arcTo. It's maybe the best way to adjust the spec.
Comment on attachment 28720 [details] silent return for r<0 We should match the spec. Even if that means the spec should be changed here. This change would also require a test case (preferably a new js-style test). I suggest you add a test to check our behavior regardless. Then at least we know when we change our behavior later to match any changes in the spec.
Comment on attachment 28720 [details] silent return for r<0 dirk and i already discussed this -- he's going to email whatwg's mailing list to get the spec changed.
Comment on attachment 28720 [details] silent return for r<0 Oh, needs a layout test :D
Sorry for my absence. I'll upload a layout test in the next days.
Comment on attachment 28720 [details] silent return for r<0 Marking as r- due to lack of layout test.
Ian Hixie denied the change of the spec. (for the moment ;-)) http://www.nabble.com/Canvas---Exception-on-arc-with-negative-radius-td22748785.html
Opened a bug in bugzilla.mozilla.org (https://bugzilla.mozilla.org/show_bug.cgi?id=491150). Let's see their opinion.
*** This bug has been marked as a duplicate of bug 37176 ***