Wouter Verhelst dixit:
>I've been thinking that rather than trying to build everything, we could
>prioritize based on what we think would be more useful to build (or
>based on feedback from users, what they would want). We could use this
>build-priority value to reorder packages in that manner. Thoughts?
This sounds better than the suggestion to effectively fork Debian
and remove some packages totally from it.
>Occasionally, you should have a look at
>http://buildd.debian-ports.org/stats/ some time. The curve is going
>steeply uphill these days, for the first time in a very long time.
>Having said that, we could still use some extra buildd hosts;
I’m already in discussion with a friend who’s also firewall and
mail admin at work, to see whether I can’t turn one, or possibly
two even, of the four VMs on my workstation at work into a buildd.
They run on my workstation, which means: they use up “spare CPU
and RAM resources” for when I don’t need them (I do occasionally
run the one or other x86 VM in Linux-KVM or VMware on that box,
but it’s got four CPU cores and 8 GiB RAM, so there’s plenty of
spare), and they may be turned off occasionally without notice,
usually cleanly though.
But they’re fast (about 200 BogoMIPS!) and available, plus almost
everything for them is already set up – so, why not? (Additionally,
I found out that having a buildd is faster than scheduling builds
and picking them up afterwards manually: ara5 built 11 packages or
so in its first *night*, i.e. half-day, alone…)
I want one of these. They cost 720 € though… good they don’t have the HD hole,
which indicates 3½″ floppies with double capacity… still. A tad too much, atm.