[BUG] bzr locks itself when pushing from branch to a checkout in the same shared repo

[Aaron Bentley]

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

Robert Collins wrote:
> So the basic condition is 
> 'bzr does not know in method A about a branch open from method B'.
> 
> But in fact, its entirely appropriate that bzr know that the master
> branch is in the same repository as the local branch, because we dont
> want to try to push the revisions twice - only once.

My bug was not with heavyweight checkouts, it was with branches.

> So why cant we just check for that - that branch.repository ==
> master_branch.repository ?

Well, we don't really have a concept of equality (or identity) for
repositories-- the closest we have is 'same location', but that doesn't
address aliasing issues.

Where possible, I think the best solution is to take all write locks
before taking any read locks.  But perhaps we should be taking a closer
look at notions like lock tokens.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFyTsQ0F+nu1YWqI0RAm4rAJ9aWj4dFaUgM4769HjVFKUwP3eYpQCfYqUF
8J/mo9lCcSvGe0AB3A/8ZR8=
=SOa4
-----END PGP SIGNATURE-----