More hacking. I've done installs across seven platforms today (okay, I don't know that I did one of them today, but I did it yesterday). And I tested kernels across seven platforms today. Things are starting to fall into place there, but the esoteric arches still have some work to do on them.
iSeries installs are hanging with anaconda in a D state when package installation completes. strace and gdb fail to attach obviously — I should try attaching them sooner. Adding debugging logs of rpm transaction bits makes the problem go away. Just adding debugging logs to tell me where I am doesn't seem to. Ideas for tomorrow: a) attach gdb early so that I can try to get a backtrace with the process already attached b) see if it happens in GUI mode c) see if it happens with the logging when I'm not tailing the log an d) get full strace output for an install. In that order I think.
Also poking at various other installer things. I erased my whiteboard yesterday (keeping the shortish list of stuff that still needed doing). The list today is back to being absurdly long. And I've yet to get to the stuff I was hoping to get to this week. And tomorrow's Friday. Blah.
Parents are coming up Saturday afternoon/evening to take me to dinner for my birthday. Other than that, I think I'm going to try to get the buildinstall changes I wanted done. Maybe move on to the more fun stuff if I get through that. Tomorrow, have to wake up at the butt-crack of dawn to go to this breakfast with the new (well, not so new now I guess) executive director at NCSSM. Should be interesting, although waking up is going to suck. I should probably get to bed so that I have a chance of actually getting up in the morning.