Comment on attachment 458562 [details] [diff] [review] fix >+<embed type="application/x-test" style="width:200px; height:400px;"></embed> >+<embed type="application/x-test" id="p" >+ style="position:fixed; width:200px; height:200px;"></embed> Presumably you intended position fixed on both the reference and the test?

Actually I intended *not* to have position:fixed on the reference. They get the same position thanks to left/top being auto.

So I guess there are some other places that got view invalidation but now need frame invalidation?

Probably. I fixed a few in bug 564991 itself.

This is what I see before patches. The black box lines up with the plugin and transparency of the from left to right of the plugin area which i am able to see the background app's left and right vertical borders through firefox.

Actually I suspect this is more of a bug in the test plugin than anything else. When a windowless plugin's rendering depends on the size of its window, and it detects a change in the window size, it should invalidate itself, right? I don't know of any promise that we make to invalidate a plugin in such a situation. The test plugin doesn't invalidate itself on window size change, but it should.

I think I'm fixing a genuine bug in the test plugin here: in the default drawing mode, when the window size changes we need to redraw the entire window since the text+borders we draw everywhere depends on the window size. (For the solid color mode, this is not true.) Please correct me if I'm wrong.

Comment on attachment 461975 [details] [diff] [review] better fix Fix to test plugin only. No risk.

The testcase plugin certainly looks fixed now. I tried to resize it a few times and it does show issues anymore. Mozilla/5.0 (Windows NT 6.1; rv:2.0b4pre) Gecko/20100808 Minefield/4.0b4pre ID:20100808124602

This bug's testcase seems to be sporadically failing on tinderbox a few times a day -- see bug 611164.