[MERGE] Add BzrDir.get_branch_stacked_on_url

[Aaron Bentley]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Robert Collins wrote:
> On Mon, 2008-09-01 at 20:21 -0400, Aaron Bentley wrote:
>> Well, I'm not opposed to this as a stop-gap as long as you file a bug
>> that we shouldn't open fallback branches until we know that we need them.
> 
> stacked_on_branches are always needed today, for the repository, branch
> history etc to operate correctly. The repository has no way to call-back
> to the branch to ask for other locations, and I'm not sure that would be
> a good way to construct it anyhow.

We need to know where to *find* them, sure.  I don't agree that we need
to open them.  There are some queries that don't require stacked-on
branches.  As we turn stacked branches into shallow branches, we will be
avoiding opening branches anyway.

>> The patch as it stands assumes too much about the way branches are
>> configured.  This would fail spectacularly with a Subversion branch, if
>> they supported stacking.
> 
> I agree that the BzrDirMetaDir implementation should indirect through
> the format object of the branch though I will note that bzr-svn branches
> are not a problem: subversion branches have their own BzrDir
> implementation which means they already have an indirection point.

IIRC, bzr-svn's control directory inherits from BzrMetaDir, so they
would be affected.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFIvMNn0F+nu1YWqI0RArF4AJ9U5hKLEvFCD6aeXr2QB4yw65mgpwCeJNkG
ejuFg9E6Nnplztkw9Atx0Js=
=I70a
-----END PGP SIGNATURE-----