Re: Moving to git

[Thomas Schwinge]

Hello!

On Tue, Mar 10, 2009 at 04:33:51AM +0100, olafBuddenhagen@gmx.net wrote:
> On Sat, Feb 28, 2009 at 06:11:07PM +0100, Thomas Schwinge wrote:
> > And I'm seeing the same (I think) problems: converting gnumach with
> > git-cvsimport will yield an unusable gnumach-1-branch -- on which you
> > would still find all the oskit files that are not present in the CVS
> > branch. [...]
> 
> Have you made any progress with that?

I threw parsecvs into the pool of possible tools to use.  From a quick
glance, it gave the best results so far, but unfortunately also exhibits
one major problem: it gets the branch-off of gnumach-1-branch from HEAD
too late, thus on the first commit on the then-new gnumach-1-branch, all
files that have been added on HEAD in the mean time (including the huge
merge-in of oskit-branch -- the reason for creating the
gnumach-1-branch).  Talking to Keith Packard (parsecvs' author), he also
didn't have an immediate explanation -- it might be due to CVS repository
corruption.  Also, the problems that the other conversion tools are
having might be due to that.  Either I get this potential error located
and corrected in the RCS *,v files, or perhaps I'm simply -- as it's a
one-time-only thing -- able to remodel a to-git converted repository into
the right shape manually.


> I'm not sure whether you saw my note on IRC: I suggest converting the
> Hurd repository only for now, and postpone gnumach until we have a
> solution for this problem. They are independent enough that we don't
> need to delay one because of the other...

I agree.


> Potential GSoC students will (hopefully) start popping up in a couple of
> days, and I really want to be able to point them to the Git repository
> right away -- at least for the Hurd itself...

I have some university work to be done right now, but I'll try to
schedule some time for this later today.


Regards,
 Thomas

signature.asc
Description: Digital signature