LNX-BBC
Our build developed some errors at the beginning of the week, but I think people managed to straighten them out. But the week turned out to be very turbulent for the LNX-BBC project. We moved to new hardware -- a very fast AMD Athlon -- for our build server, and Nick started to experiment with ccache, a "compiler cache" program which remembers the results of compiling particular sources, so that builds become faster after they've been run once.
As you might imagine, ccache is by Tridge. Go Tridge!
The new machine made builds (which used to take about eight hours) complete in two or three hours. The new machine together with ccache made builds complete in what seemed to be under an hour, which was very impressive (because it means that it would actually be possible to test the results of a particular small change by waiting for a new ISO image and burning it to CD).
Unfortunately, the new machine has turned out to be very unreliable; its memory becomes corrupt and we get intermittent compile failures, often with a sig11 error (something I hadn't actually experienced since I had some bad RAM in my old 486 machine). So the new server is down now, pending the availability of some hardware which will work reliably.
In the meantime, we managed to fix a number of bugs. I'm currently working on a system to allow various things to be downloaded from the network and to have their signatures checked with GPG. This is a part of having a better and more secure downloadable package infrastructure, and also allowing people to create custom BBCs more easily.
So I'm writing a shell script and trying to think of every possible case, including cases we don't use ourselves.
