[MERGE] StackableBranch

[Aaron Bentley]

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

Colin D Bennett wrote:
> On Tue, 10 Jun 2008 01:53:00 -0400
> Aaron Bentley <aaron at aaronbentley.com> wrote:
> 
>> Ian Clatworthy wrote:
>>> # Bazaar merge directive format 2 (Bazaar 0.90)
>>> # revision_id: ian.clatworthy at canonical.com-20080610045325-\
>>> #   hi3yug3nasoe6e1h
>>> # target_branch: ../Development1/
>>> # testament_sha1: faebafc1e65c0665e9bf55332d74bbfdf373afa0
>>> # timestamp: 2008-06-10 15:17:34 +1000
>>> # base_revision_id: ian.clatworthy at canonical.com-20080610013832-\
>>> #   9dt6c5h67eckiezo
>> Please don't submit unmergeable merge directives.
>>
>> Bundle Buggy is freaking out because it doesn't have a copy of
>> ian.clatworthy at canonical.com-20080610013832-9dt6c5h67eckiezo and it
>> can't get one from ../Development1
>>
>> Please either use bzr.dev as the submit branch, or supply some other
>> valid external URL for the submit branch.
> 
> Doesn't the "revision bundle" within the merge directive contain all
> the necessary information to merge the merge directive?

Not if you use the wrong submit branch when you generate the merge
directive.  If you use a submit branch containing revisions that no one
else has, you're tricking send into generating a bundle that does not
provide the revision data necessary to merge it.

> I thought that
> the revision bundle made the public branch/submit branch unnecessary.

The submit branch is a parameter to the "bzr send" command.  I think
you're talking about the target_branch value field in a generated merge
directive.

> I could imaging BB using the submit branch to determine if this is
> something that it should consider for merging to bzr.dev or not, but as
> far as actually merging the merge directive into a branch of bzr.dev,
> shouldn't it work fine without a submit branch?

Assuming you're talking about "target_branch", it isn't used except to
fetch revisions not present in the bundle.  This could happen if you
were merging into a local copy that was out-of-date.

However, it would be a format violation for a merge directive to not
include a target_branch field.

> I'm just trying to clarify my understanding of how 'bzr send' and merge
> directives work.  I really love 'bzr send' now that I've discovered
> it and learned how to use it.

I'm glad.

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

iD8DBQFIT90j0F+nu1YWqI0RAtMBAJ4tKlaIA2IgNiWorwgj2WpP0XUV6QCfXZpG
s1DxQawDAhNRsJ2lTWaaY0g=
=R4p8
-----END PGP SIGNATURE-----