If a non-accelerated canvas2D call is made (e.g., strokeRect) followed by a call to drawImage() of that canvas to another canvas context, the results will be incorrect (the stroke is not shown).
This can be seen on sites such as http://canvasrider.com/tracks/695376, although there doesn't seem to be a layout test that covers it.
Created attachment 89061 [details]
Comment on attachment 89061 [details]
View in context: https://bugs.webkit.org/attachment.cgi?id=89061&action=review
> +<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML//EN">
just <!DOCTYPE html> should suffice
Landed as http://trac.webkit.org/changeset/83582.