[ANN] bzr-colo 0.0.1: colocated branches using present technology

[John Arbash Meinel]

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


...
> Oops, my local branch has now been bypassed and switch has become a 
> lightweight checkout of the remote branch. The result is that operations 
> within switch are as slow as could be when the remote branch is somewhere 
> across the WAN (which in my case it is in the real world).
> 
> So that's the crux of my problem: why do these two produce different results, 
> and should they?
> 
> bzr switch trunk
> bzr switch ../trunk
> 

It is because of how the 'trunk' short name is getting looked up. If you
look at bzrlib.builtins.cmd_switch._get_branch_location

It first checks to see if this is a bound branch, before checking if it
is a branch reference. And use the master location as the base for
determining the relative path.

Which is inconsistent with bzrlib.switch._set_branch_location, which
first checks to see if this is a lightweight checkout before checking to
see if it has a master branch.

Probably "just a bug" and we should switch _get_branch_location to check
for a lightweight checkout first.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAktLqSsACgkQJdeBCYSNAAMUsACgjo43r4VfGCs+xaUnL+YDvpJi
Y5wAoKy0DPQj8hFW5vgIfe70aigZqtd6
=AHcR
-----END PGP SIGNATURE-----