svg/custom/use-css-events.svg seems to have regressed on feature-branch Not sure when or how. But actual and expected images seem to disagree.
The test case is still rather complicated. Not entirely sure what it's testing.
Created attachment 17656 [details] Initial patch Also fixes several existing other test cases (all regresssions).
Comment on attachment 17656 [details] Initial patch r=me
Landed in r28325.
Comment on attachment 17656 [details] Initial patch Um, it's seems kinda lame that any code depends on inner SVG's having width() height() set on their renderer. width()/height() are HTML rendering tree concepts. I guess they can be re-used in this manner.