Branches without working trees

[Aaron Bentley]

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

Robert Collins wrote:
> Some differences here - it didn't make sense to me that the push
> location would be preserved by branching

This doesn't logically follow.  In my experience, there are two useful
kinds of branching.

cp -r, aka branch cloning -- This produces an exact copy of the branch,
with the same settings

bzr branch, aka smart branching -- This produces a branch whose history
is the same, but with smart updates to its settings.

When I clone a branch, I definitely do want the pull location preserved.
 Stands to reason I want the push location preserved, too.

When I do a smart branch, it makes sense for the old branch to become
the parent of the new branch.  By analogy, the old branch should be the
default push location of the new branch.

, so I stashed the push location
> in ~/.bazaar/branches.conf.

I really am not fond of this idea.  I would like to associate the
push/pull/submit/child-submit/merge-partner data with the tree, not its
name or location.  That gives us the right behaviour for branch cloning,
and has no ill effect on smart branching.  It also provides better
information-sharing.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDZQA80F+nu1YWqI0RAnNcAJ4lAVPL6kiQcFzv7n9HxISj4hkvagCfWCRS
MB+R2NhFRrXEJd4XLV0hvyA=
=QBqd
-----END PGP SIGNATURE-----