* On 09 Apr 2013, Brendan Cully wrote: > I think Derek is right, we should be handing out the commit bit a > little more freely. I think DGC proposed once we use something like > Mercurial's crew repository as a staging area for patches that we > think are good but need review before applying to the main repo. This > might make it easier for new committers.
Indeed I did, and I agree it would help especially for enlivening the occasional developer community. I don't have strong opinions on whether that should be at mutt.org or at bitbucket -- or whether the official repo should either, for that matter, but maybe that should be considered. I think the point is well made that vcs hosting community platforms like bitbucket and github add traction, and I'm using bitbucket for most new work these days so it's not really a hurdle to jump anymore. Also, as much as I like Trac, Bitbucket's issue manager is really not bad, may require less administrative effort, and improves the overall -apparent- level of integration. > The other big thing that mutt gets blocked on is bug fixes and the > effort it takes to avoid regressions. This takes a lot more time than > it should because we don't have any kind of test suite. Rocco once > started working on one at https://bitbucket.org/pdmef/mutt-testdrive > but I don't know how far he was able to get with it. This would be a > great project for someone with a bit of time, with I think very large > benefits for everyone: applying patches would be safer, and reviewing > them would take less work, so we'd be able to get a lot more through > our limited maintenance bandwidth. Agreed: regressions are a big problem and it would be great if someone could implement tests, but it's also a difficult thing to do with a tool that's so interactively driven, and inherently limited. It might also help to begin applying a prettifier during the pre-commit or pre-push code check. Configuration params for the prettifier in repo, of course; and perhaps with a target in the makefile: "make pretty". Indent is probably too variable for this; there are at least three incompatible versions. I've used uncrustify, but would be open to anything portable, well-versioned, and unforked. -- David Champion • d...@bikeshed.us
