The continuing repository-collision saga

[Ben Finney]

Howdy all,

My long-suffering collaborator, Rene Engelhard, got some assistance
from the Bazaar IRC channel regarding the latest patch bundle I'd sent
to him.

=====
19:40 < _rene_> jelmer: ping?
19:40 < _rene_> jelmer: how do I upgrade the remote copy on bzr.d.o to the "new" format?
19:41 < _rene_> jelmer: because now, when I try to commit, it complains about "different rich-root support"
[…]
20:08 < jelmer> _rene_: you're looking for a format that allows that bundle *and* is supported by sid?
20:08 < jelmer> s/sid/lenny/
[…]
20:08 < _rene_> I want a solution which lets me bzr merge those bundles he sends and which works in sid *and* works with bzr.d.o
20:09 < jelmer> _rene_: try upgrading to --rich-root-pack
20:09 < jelmer> _rene_: that's supported by lenny, and will work with that bundle
20:09 < jelmer> _rene_: downgrading from --1.9-rich-root is also possible
20:09 < _rene_> how?
20:09 < jelmer> _rene_: "bzr upgrade --rich-root-pack"
20:10 < _rene_> nevermind, already rm -rf'ed debian
20:10 < _rene_> and that will also work with bzr.d.o as-is?
20:10 < jelmer> _rene_: yes
20:10 < _rene_> or do I need some action on it, too?
20:10 -!- sabdfl1 [~sabdfl at 165.145.86.85] has quit [Ping timeout: 480 seconds]
20:11 < jelmer> _rene_: If there's an existing branch on bzr.debian.org, you'll have to upgrade that to rich-root-pack as well
20:11 < peterS> eh, alioth is running lenny now?
20:11 < _rene_> jelmer: how?
20:11 < _rene_> jelmer: I doubt alioth runs lenny.
20:11 < peterS> last I checked it was on etch
20:11 < jelmer> Sorry, it might not
20:11 < _rene_> jelmer: and yes, I am always talking about existing branches
20:11 < jelmer> but it's got the bzr from lenny, that must be what confused me
20:11 < peterS> ahh
20:12 < _rene_> jelmer: otherwise I would not have the problem of merging something to that branch :)
20:12 < _rene_> any OOo version switch and even distro-switch gets a new branch :)
20:13 < _rene_> jelmer: so I should run bzr upgrade on bzr.d.o?
20:13 < jelmer> _rene_: yep
20:13 -!- bono (bono) [bono at 118-160-168-40.dynamic.hinet.net] has joined #debian-devel
20:13  * _rene_ sighs
20:14 < jelmer> _rene_: Alternatively, you could ask Ben for a bundle that doesn't use rich roots
20:14 < V13> I guess... the answer to my question is 'no'...
20:15 < jelmer> _rene_: (rich roots are a feature required for bzr-svn and nested trees in bzr, there's no particular reason for using them here)
20:15 < _rene_> ok, will do that then
=====

But I've taken explicit steps to ensure that I'm *not* using a
rich-root repository for this branch. The repository is definitely not
rich-root:

=====
$ bzr info $HOME/Projects/debian/openoffice.org/
Shared repository with trees (format: pack-0.92)
Location:
  shared repository: /home/bignose/Projects/debian/openoffice.org
=====

The branch was created in that repository. I don't quite understand
the info output:

=====
$ cd $HOME/Projects/debian/openoffice.org/debian.devel/
$ bzr info
Repository tree (format: unnamed)
Location:
  shared repository: /home/bignose/Projects/debian/openoffice.org
  repository branch: .

Related branches:
  parent branch: /home/bignose/Projects/debian/openoffice.org/debian.unstable
  submit branch: http://bzr.debian.org/pkg-openoffice/packages/openofficeorg/3.0.1/unstable/
=====

What's going on here? When I sent Rene a previous patch bundle using
Bazaar 1.5, it applied cleanly against Rene's repository. The only
difference AFAICT is that I'm now using Bazaar 1.13rc1. I have *not*
changed the repository or branch formats in between.

-- 
 \     “Don't be afraid of missing opportunities. Behind every failure |
  `\         is an opportunity somebody wishes they had missed.” —Jane |
_o__)                                          Wagner, via Lily Tomlin |
Ben Finney