Open Embedded build experiment
As I am writing this, I try to build 'nano' for the Beagle Board on my laptop, using the Open Embedded build chain.
So I got my machine up and running again, which was more of a struggle than I thought. My laptop is dual-boot Windows/Debian, so the obvious thing to do would have been to access /etc/sysctl.conf from Windows and undo yesterday's fatal changes. However, the ext driver for Windows does not support an inode size of 256, which is the inode size of the ext file system that my Linux runs on. So I had to go for plan B, booting from a Linux live CD, which of course I didn't have with me in Banff. So I spent a long time waiting for Ubuntu ISO image download, and guess what: it wouldn't boot. When, after another long wait for a download of the System Rescue CD, that didn't boot either, I almost got worried until I remembered the acpi=off boot option, which has saved my life at several occasions already. It worked (and probably would have with the Ubuntu CD as well), so I undid my changes to the kernel settings and got back a booting Linux system.
More interesting to probably both me and you is the fact that I then continued to set up my Open Embedded build environment. I am right now building the nano text editor for Open Embedded, which takes a while, but at least I haven't got any error messages so far (knock on wood). Should this work, I'll try to transfer the built files to the Beagle Board and try to get nano running on it. If that turns out ok, I'm gonna try to build Pure Data for the Beagle Board. And if my residency is still not over by then, why not custom-build my own Angstrom image?