StGIT: stgitformatversion vs. stgit.formatversion

What about using branch.*.stgit.formatversion to store the version, to
be consistent with other stgit-specific per-branch settings ?

Bet regards,
-- 
Yann.

On 2007-06-02 21:16:41 +0200, Yann Dirson wrote:

What about using branch.*.stgit.formatversion to store the version,
to be consistent with other stgit-specific per-branch settings ?

Good idea. But we should probably fall back to stgitformatversion (or
better, upgrade from it) to be nice to those running Catalin's master.

It happens a bit too often that we find problems with patches only
after Catalin has published them on "master". (At least that seems to
be the case with patches coming from me!) Maybe a "pu" branch
(maintained either by Catalin or someone else) would be a good way to
smoke them out before they're written in stone and distributed to lots
of people.

If the stgitformatversion patch had been on such a branch, we could
just have edited it and not have to worry about upgrading old configs
(aside from posting a mail on how to upgrade manually).

-- 
Karl Hasselström, kha@treskal.com
      www.treskal.com/kalle

On 03/06/07, Karl Hasselström [off-list ref] wrote:

On 2007-06-02 21:16:41 +0200, Yann Dirson wrote:

quoted

What about using branch.*.stgit.formatversion to store the version,
to be consistent with other stgit-specific per-branch settings ?

I agree as well.

Good idea. But we should probably fall back to stgitformatversion (or
better, upgrade from it) to be nice to those running Catalin's master.

It could, indeed, fall back to stgitformatversion and upgrade from it.
Anyway, I'm not sure there are that many running the master branch but
this would be needed for us as well.

It happens a bit too often that we find problems with patches only
after Catalin has published them on "master". (At least that seems to
be the case with patches coming from me!) Maybe a "pu" branch
(maintained either by Catalin or someone else) would be a good way to
smoke them out before they're written in stone and distributed to lots
of people.

Well, I consider "master" to be a development branch anyway. It would
be useful to get an idea of how many are using this branch (people not
following the GIT list are probably using the stable releases anyway).
A "pu" branch would make sense for more experimental stuff, like the
DAG patches.

If the stgitformatversion patch had been on such a branch, we could
just have edited it and not have to worry about upgrading old configs
(aside from posting a mail on how to upgrade manually).

But now I have plenty of branches in my Linux tree already, so I
wouldn't do it manually. We already have checks for the repository
version, adding this would probably only be 2-3 lines.

-- 
Catalin

On 03/06/07, Karl Hasselström [off-list ref] wrote:

It happens a bit too often that we find problems with patches only
after Catalin has published them on "master". (At least that seems to
be the case with patches coming from me!) Maybe a "pu" branch
(maintained either by Catalin or someone else) would be a good way to
smoke them out before they're written in stone and distributed to lots
of people.

BTW, if any of you is willing to maintain such a branch, I'll happily
pull from it when it gets stable.

-- 
Catalin

On 2007-06-07 22:55:09 +0100, Catalin Marinas wrote:

BTW, if any of you is willing to maintain such a branch, I'll
happily pull from it when it gets stable.

I'm on vacation the next few weeks (though this week I'm at the FCRC
in San Diego if someone wants to say hi!) and have rather spotty
internet connectivity, so I haven't really been able to follow the
mailing list. But I think this would be useful, so I'll probably start
doing it as soon as I'm all settled again.

-- 
Karl Hasselström, kha@treskal.com
      www.treskal.com/kalle