© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
0e1f616
(Get the code!)
I also have this problem and have tried a few things to narrow down what causes it.
We have a central server with a shared repository that holds quite a few branches.
There seems to be just one branch in our repository that has this problem. Most other branches I have tried are fine.
For that branch bzr branch and bzr checkout fail in this way from where ever I try.
I have tried normal remote access to our server.
I have also taken a local filesystem type copy of the repository onto my development machine and tried local branching from outside the repository and branching within the repository with various bzr versions.
These tests have been tried with bzr 2.3.4, 2.4.2 and 2.5.1. Version 2.3.4 always works, the other two always fail on this branch in particular. They all work in all scenarios on other branches in the same repository.
I can create a new shared repository and branch this troublesome branch into it using 2.3.4. I can not then branch or check out from that branch in the new repository with the newer versions of bzr.
Interestingly, I can sit in any of the repositories and do bzr init foo; cd foo; bzr pull bad-branch and that works without errors with any revision. However, I then cannot branch from that foo branch.
I have run bzr check and bzr reconcile with no improvement. I have done this on the copy of our server repo, the new repo mentioned above and my usual development shared repo on my machine and tried these tests again afterwards .
The branch that fails has a long history. It was originally created in I think a 0.92 format repo with rich-roots. The branch was brought in using bzr-svn. The repository has been upgraded twice. To the 1.14-rich-root format and more recently to 2a.
I have tried a superficial comparision of the dirstate module in 2.3.4 and 2.5.1 and can't see any real difference in the code around the area that throws the exception.
I am wondering if the repository data for the troublesome branch is damaged somehow and the damage is being blindly copied around by the older bzr version, but the newer ones complain.
Is there anything I can do to investigate further?
Many thanks