Equivalent to svn tags?
Aaron Bentley
aaron.bentley at utoronto.ca
Thu Oct 25 17:08:59 BST 2007
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Matthew D. Fuller wrote:
> On Thu, Oct 25, 2007 at 10:25:57AM -0400 I heard the voice of
> Aaron Bentley, and lo! it spake thus:
> I see that as a social issue; managing what's in a group of branches,
> rather than a denial of the usefulness of the branch groups. (Note
> that I don't necessarily endorse cramming the concepts of 'branch
> group' and 'repo' together; that may turn out to be the _simplest_
> thing, but it's not necessarily the best).
Well, I guess I wouldn't stand in the way of such an approach, but I'd
definitely want to be able to opt out of it. I'm happy to use the
internet when I want access to old branches, especially if all the
releases are tagged.
>> So this is exactly what tags are useful for-- they provide a way to
>> reference particular releases, without
>
> But I'm not necessarily interested in particular revisions. I'm
> interested in branches.
Could you say a bit more about why, then?
When you were discussing bzr.dev, you said:
> If older releases were just a 'co' of their release branch away, I'd
> test them too.
I don't see why you'd want to test revisions other than the release
revision, and it's certainly easy to check out a tag.
> This isn't svn; we KNOW that branches and
> tags are two different things, and both have uses. And using trickery
> to insert revs into your ancestry that don't belong, just to make a
> revision visible, is definitely a hack.
I *think* this is a reference to John's:
>> bzr merge ../branch-0.3.0
>> # Revert any 0.3.0 specific changes, like the version properties
>> bzr commit -m "Merging branch-0.3.0 into trunk"
I strongly recommend doing that irrespective of tagging. So for me, it
wouldn't be "just to make a revision visible", thus not a hack.
> This sort of thing doesn't show up in bzr because we don't _have_ any
> long-lived public branches.
What about nested trees?
http://code.aaronbentley.com/bzr/bzrrepo/nested-trees
I think this qualifies as public, because other people have used it.
And it's certainly long-lived.
> It would almost be
> tautological that I'd want to mirror the whole thing, since the
> criterion would be pretty much "would this branch be generally useful
> to the community".
Well, I still think that you're not giving enough credit to individual
preferences.
> I'm confident that in that sort of community, walking up and offering
> a VCS by saying "Look, you have to individually and explicitly name
> every upstream branch you want to track"
Who's saying that? It's incredibly easy to mirror
http://bazaar-vcs.org/bzr/. It's just rsync instead of cvsup.
Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFHIL+b0F+nu1YWqI0RAsEgAJ9oBzz5hSelGkKR1iNAj3C7G9w6YgCbBIQg
i0Ta4fiaEd0Yy6zg+xq98i8=
=Tk57
-----END PGP SIGNATURE-----
More information about the bazaar
mailing list