On Mon, 14 Jul 2008, Roman Zippel wrote:
On Wed, 2 Jul 2008, Geert Uytterhoeven wrote:
On Wed, 2 Jul 2008, Finn Thain wrote:
Perhaps this is a good time to move to a git tree on kernel.org?
Perhaps...
Thanks to Ralf Baechle, I have a git conversion of Linux/m68k CVS of 4
months ago. But I still don't know what's the best way to proceed.
I already have an account on kernel.org, so I've just put my patch queue
there. It should show up soon at
http://www.kernel.org/pub/linux/kernel/people/geert/
I'm not sure if standard conversion tools properly recreate our repository
structure, especially one has to know what that the tags mean to recreate
the merge points.
Over the weekend I played a little a script to extract the history
information as patch series from all the linux-* to the corresponding
m68k-* tag. I have to make some small changes to the repository to make
it easier for the script to properly detect all changesets (e.g. it has
small problems with duplicated logs, as they are used as index), but some
tag information need fixing too, as there are a few commits done between
tagging and import, so they get associated with the wrong release.
Assuming we want to preserve our history in a git tree I need two things.
First, I have an older copy of the repository, which I used for testing
and it mostly works, so a I need a copy of the last repository and it
shouldn't take too long to create the patches (most of the time will be
testing that the patch series really is equal to what has been tagged :) ).
Second, I could use some help on how to merge this into git, as I'm not
using git for much more than fetching the latest sources. :)
One problem here that the current git only starts at 2.6.12, where our CVS
starts at 2.4.12. We could create two repositories, where the first is a
simple CVS conversion as historical tree and the second is based on Linus'
tree starting from 2.6.12, which we can use as our development repository.
Has anyone already something like this? I know it's possible, but I had to
dive deeper into the git docs to find out how. If someone has example
scripts, it might really help and save some time. We basically need to
create a m68k branch, the patches are imported to it using the existing
v2.6.x tags, then the merge needs to be redone. The first patch in a
series is the merge patch, which produces the post CVS merge, so the
merge needs to be synced to this merge patch and committed.
Any other ideas?
The main question is now if and how we want to preserve the CVS history.
IMO it's possible to preserve it, but if nobody needs it, I don't need to
spend more time on it...
You probably want to talk to Ralf, he has some experience with this...
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
--
To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
