On Tuesday 03 February 2004 14:43, [EMAIL PROTECTED] wrote: > > I understand why a keyword might be useful in marking ebuilds that'll go > > into a stable branch. However, you haven't ansewred my question - why > > have a separate cvs branch at all? Why not just use keywords as we do now > > for arch/~arch? > > My own opinion: > > 1. The size of the tree. If this is to be an enterprise tree, its likely > that it won't have an ebuild for every package as most enterprise people > won't want x (name an obscure desktop related package). Quicker > synchrosing and easier backup/restore and less clutter is good. This > isn't a big issue really though.
For snapshots we can easily filter a tree on keywords to produce a smaller, cleaner one. I've also heard (don't remember where) people talk about filtering for your ACCEPT_KEYWORDS in emerge sync. That would leave stuff in files/ which doesn't clearly belong to a specific ebuild file, but the total size of such files isn't too big. And, right now I'm thinking about the developer end of things. It'd be more comfortable for me to work with a single cvs tree and keywords. Fixing up the rsync end is an implementation detail :-) > > 2. Guaranteed someone will "clean" an ebuild thinking its to old and > needs to be removed. This is a big issue. > > If its a seperate tree where only people who know the rules of the tree > work, its more likely to stay consistant and not have stuff removed that > shouldn't be. Only the people who know the rules of our main tree should work there, too. Including the rules about not deleting whatever shouldn't be deleted. Please see my reply about this to trance. -- Dan Armak Gentoo Linux developer (KDE) Matan, Israel Public GPG key: http://dev.gentoo.org/~danarmak/danarmak-gpg-public.key Fingerprint: DD70 DBF9 E3D4 6CB9 2FDD 0069 508D 9143 8D5F 8951
pgp00000.pgp
Description: signature
