Luke Kanies wrote: >> I'm not sure that we have the same definition of stable. For me >> stable means doesn't move forward, has no new features but only >> critical bug fixes (ie the debian definition of stable). And I'm >> not sure it's a good idea to have people committing against this >> branch, we'll risk to have no new features ever :-) > > It's true that my definition of stable is a bit skewed, and we > should be leaning more toward the Debian definition. Good point.
Perhaps using the terminology as git.git does would work? In that case, 'maint' is what would be more like Debian's definition of a 'stable' branch. The 'master' branch would be where things that had cooked a bit would go, and where many folks would base their work. More active development would go on in 'next'. You can't please everyone by default, of course. But for new developers, the best place to start is likely not on large, backward-incompatible features, but on fixing up existing issues, which would be based on master. One important aspect of the git.git model is that changes are merged upwards as much as possible, e.g. maint -> master -> next. So if a fix is needed that applies to the 'stable' (i.e. maint) tree, that's where it should be applied first. Then maint can be safely merged to master, and so on. When there are large changes, it would probably make sense to have a maint-0.24, maint-0.25, etc. Junio describes this better than I could, and includes some examples of the commands used in the git MaintNotes document: http://git.kernel.org/?p=git/git.git;a=blob;f=MaintNotes;hb=todo#l80 (As always, consider this for what little it's worth, as I'm not much more than a puppet packager and user.) -- Todd OpenPGP -> KeyID: 0xBEAF0CE3 | URL: www.pobox.com/~tmz/pgp ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Sometimes I think I understand everything, then I regain consciousness.
pgpcS6hQGhELY.pgp
Description: PGP signature
