There was nothing bad about Tuesday, but it wasn’t all-out “awesome”. I went to the Abstract Interpretation reading group meeting in the morning; my dad came and took me back to his house where I set up his DSL; and then he brought me back and I worked on Xlib/XCB some more. I had tried to run the X Test Suite and x11perf on my laptop overnight but x11perf got a BadMatch error during the PutImage tests (turns out it doesn’t work on a 640x480 display) so I had to restart most of the testing in the morning.

By evening I had complete results. Xlib when running on XCB passes one test that X.org 6.8.2 Xlib doesn’t, and otherwise the two have the same failures. Further, performance is comparable: there are some x11perf tests that get much lower throughput on XCB, but a) we’ll fix that, and b) I think the issue (whatever it turns out to be) won’t affect real-world performance. I’m at least not noticing any performance problems now that I’m using Xlib/XCB for everything on my laptop.

Shortly after midnight we got slides off to Keith, with a pretty graph illustrating the performance numbers and stuff. I should put that up on the web somewhere.