Re: [PATCH/RFC] git-svn: Fix branch detection when repository root is inaccessible

[Mattias Nissler <mattias.nissler@gmx.de>]

Hi Eric,

On Mon, 2009-07-06 at 14:27 -0700, Eric Wong wrote:
> Fixing the minimize logic is definitely a wanted improvement I haven't
> been able to fix myself.  Unfortunately, your patch breaks the
> t9138-git-svn-multiple-branches.sh test for me.  

I see the same here, not sure why I've missed it before. I've had a look
at the problem and found that the test is actually depending on the
minimizing logic, i.e. needs to specify -d project1/<tag-or-branch-dir>
for the tag or branch destination where the project1 prefix is due to
the minimization at clone time. Dropping the project1 part makes the
test succeed. I'll simply update the test to do so.

I also think this is much better from a usability point of view since
you can now use the original branch or tag dir you've given to clone in
order to specify the branch or tag destination.

> I haven't looked at it
> at all, but I'll try to take a look at it later today after I process other
> things in my overflowing work queue; but I definitely want access issues
> fixed for people with restrictive repo setups.
> 
> One thing I would do is to split out the making of dup_changed_paths()
> the default behavior into a separate patch.  I think this is a good
> change since it makes things less surprising for future hackers of
> git svn.

Ok, will do so. I'll send updated patches in reply to this mail.

Finally, a big thanks for the nice job on git-svn. I've just experienced
lots of trouble with the mercurial SVN integration and it's really a
pain in the ass even to just get a simple clean rebase done properly.
git-svn is really way more mature and it's a pleasure to use it for
everyday work.

Mattias