Re: [tlug] Call for presenters - March 14th technical meeting

[Edward Middleton <emiddleton@example.com>]

Stephen J. Turnbull wrote:
>   Where I really need fine control over
> the history graph, I use raw git.  If I don't have a git repo, I make
> one.
>   
[snip]
> Edward Middleton writes:
>   
>  > I have found git-svn works pretty well with subversions
>  > repositories but I can't seem to find anything equivalent for
>  > Darcs.
>
> You mean git-darcs?  Or darcs-svn?
>   

Probably what you were referring to above.  I want to keep everything in
a local git repository but be able to pull and push changes to a remote
darcs repository.  This is how I use git-svn.

i.e.

# git-svn rebase

to rebase to latest server state and

# git-svn dcommit

to push my changes to subversion repo

It effectively makes the svn repository a remote branch.

Someone has made a ruby script[1] to pull remote repositories into a
local git repository and this is almost what I want but it would be good
to also be able to push changes.  I haven't really looked at darcs
enough to know whether this is possible.  Basically I don't want to have
to bother with yet another VCS.

>  > Subversion was just an incremental improvement on RCS->CVS but
>  > probably still the best for a purely centralized repository
>  > approach.
>
> Well, yes and no.  It's not the radical innovation that DVCS is, but
> Subversion does many things so much better than CVS by now that it's a
> different model.
>   

What features do you think make subversion qualify as a completely
different model?  Granted, building a versioned filesystem was a rather
different approach to implementing a VCS but it seems to be fairly
orthogonal to the VCS model.

Edward

1. http://www.sanityinc.com/articles/converting-darcs-repositories-to-git