Re: 2d speed regressions
Note that all the issues I mentioned were about correctness, not performance. We generally (I sometimes stray from this) prioritize bugs in terms of "does it prevent the system from working", "does it render incorrectly," then "does it reduce performance". We've spent a lot of time on the first two, and we've still probably got a hundred "my machine stops working" bugs to deal with. It's way more important than performance, generally.

But, Clemens, there's also the problem that the bugs you're reporting are about how performance has dropped for your Java 2D code. You chose to NIH cairo, and so you may not be seeing the benefits when we fix performance problems for cairo (what people like, you know, Firefox use. And every other app on my desktop). Your Java project is a lower priority for me to work on than "every app on my customers' desktops," so we probably won't get to them until we've got at least all the "easy" stuff for performance in OpenGL and cairo done. And I'm sure I've got "easy" stuff I'd like to do that would keep me busy for a year right now.