On Fri, Jun 10, 2011 at 2:43 PM, Matthias Rampke <matthias%rampke.de@localhost> wrote: >> ... because Dragonfly is very close to NetBSD? > > Maybe … but mostly because most of the software is pretty portable to begin > with. A lot of it is because of plain ordinary work, especially the initial work from Joerg Sonnenberger, to make more packages build on DragonFly. The build success rate for the last few quarterly pkgsrc releases has been very high - higher than it was historically, if my casual memory serves me. I've been using pbulk on DragonFly and mailing results to the pkgsrc-bulk%netbsd.org@localhost mailing list. It's been very effective in identifying problems and, even better, which package breaking has the most effect. I don't think there's necessarily a need to tell committers to use already-scarce time to test their software on new platforms. Coming from the viewpoint that the only useful test builds are ones that break, if we have regular automated bulk builds with results, the maintainers can find out about where their package fails without having to take time to test. The short version: more people doing bulk builds on whatever platform they have handy is possibly the most results for least commitment in this case.