[makkonen]

I took my own, and even went to the trouble of recalibrating right before grabbing the data. It was pretty consistently off. When I clicked at about 3mm in from any corner, it would read at the farthest point into the corner. (minimally off, but enough to be annoying). Not sure if it was just the way I calibrated it, or if they're actually using different data.

phh is still working on calibration code that can run in linux -- he doesn't like the idea of relying on winmo. At least we're starting to have options, and we know we can get it right with some tweaking.

In other news, new kernel coming -- fixes sound, at the cost of power off not working. (for some reason the power off fix from a few days ago makes the audio RPC timeout at the kernel level. really weird. I assume there's a more elegant fix, but for now, I'd rather have audio and no shutdown than vice versa)