[MERGE][bug #111702] bzr init <remote_branch> connects multiple times

[Vincent Ladeuil]

>>>>> "aaron" == Aaron Bentley <aaron.bentley at utoronto.ca> writes:

    aaron> Aaron Bentley has voted +1.
    aaron> Status is now: Semi-approved
    aaron> Comment:
    aaron> I think you are right that providing possible_transports is a
    aaron> good way of doing caching without actually doing caching.  We
    aaron> certainly don't want to wind up with two variants of every
    aaron> operation that takes a url: one with a url parameter and one with
    aaron> a transport parameter.

Great to see some consensus on the idea :-)

I have already written tests for branch, cat, checkout, init,
init-repository, merge, missing, pull and push.

That seems to cover most of the cases where we want to ensure
that only one connection is issued (any other I missed ?).

I plan to add some more for bound branches.

One thing bother me though, the medusa ftp server seems very slow
and each test takes between 2 and 4 *seconds*.

So I hesitate a bit about adding redundant tests (somme commands
rely on the same objects underneath and more tests may not cover
more code paths).

Overall, only init and branch fail to pass which is good news ;-)

     Vincent