What a grueling day of user studies - we did four 1.5-hour sessions, with suboptimal sleep. I'm just taking notes this time around, but even that requies actively thinking about what can be designated a problem and relying heavily on my short-term memory as I scribble away. I have three more days just like it, too (though hopefully I can correct the suboptimal sleep issue).
Last night I finished my SkimDecaf interpreter and pretty printer at about 2am. It was the first largish framework-based computer science class project I've enjoyed in ... well, maybe ever - even graphics involved way too much busywork and C++ pain, and let's just not talk about operating systems. This project gave immediate and gratifying feedback - you feed code in, and if it doesn't print or run the way you expect it to, it was easy to debug. It also wasn't too difficult to get a sense of what the provided skeleton code was doing and why, which is often my biggest hurdle in these kind of projects. Eclipse's interpreter-like interface was very helpful, too.