[SciPy-dev] Question for git-svn masters

Matthew Brett matthew.brett@gmail....
Wed Nov 25 15:07:36 CST 2009

Hi David, Pauli, git-svn masters,

I have used the instructions here to good effect... :


I cloned the git-svn mirror to good effect, and posted my repository
back to github:


If I'm working from the original local clone of the scipy mirror, all
works well, I can do git svn rebase and so on.

If I clone from the github mirror:

git clone git@github.com:matthew-brett/scipy-work.git

then the resulting clone has lost the relationship with SVN:

b312@tom:~/tmp/scipy-work$ git svn log
Migrating from a git-svn v1 layout...
Data from a previous version of git-svn exists, but
	(required for this version (1.6.5) of git-svn) does not exist.
Done migrating from a git-svn v1 layout
fatal: bad default revision 'refs/remotes/git-svn'

not surprisingly, because I have nothing about the svn in my .git/config

If I do this incantation (from the GitMirror page):

git-svn init -s --prefix=svn/ http://svn.scipy.org/svn/scipy
git-svn rebase -l

after a long time, I get:

Unable to determine upstream SVN information from working tree history

I tried Googling for this error, but reached a state of childish
confusion rather rapidly.  This page stood out as referring to my
problem, but I simply don't understand the issues well enough to see
an obvious fix from there:

I'm using git-1.6.5 (latest stable compile from source).

Is there a way to reassociate my clone with SVN?

Is there a way to push this reassociation to my github mirror so that
'git clone ...' brings the SVN association with it?

Thanks a lot,


More information about the SciPy-Dev mailing list