On Thu, Sep 25, 2008 at 16:12:54 -0700, Simon Michael wrote:
> I feel the pain you're running into above.

Thanks! But actually the pain above would be felt whether there was a
release branch or not (the issue is that we would need to generate a
patch to the release branch which will conflict with the unstable
branch; whether or not there is an intervening stable branch is
irrelevant)

> It's hard to keep things  
> simple. I don't understand it as you do, but I still wonder if  
> maintaining a parallel release branch is bringing you any real benefit ?

So the release branch was really to allow patches to be accepted into
stable (so that stable be closely enough in synch with unstable that it
is practical for people to use it as their main development branch).
Another goal was to allow the stable maintainer and release manager
jobs to be done by different people.

That said, I can see that this is causing some confusion, and I concede
that this has caused me some technical difficulties, namely that our
approach of running make dist on darcs.net relies on the assmuption that
the stable and release branch are the same thing.  Since release is only
a subset of stable, running make dist has the unfortunate effect of
including too many patches.  The workaround I had to use was to upload
my own tarball, run make dist, and clobber the produced tarball with my
own.

> An alternative might be to treat the stable trunk as your release  
> branch, with a code chill/freeze period right before release as some  
> well-known projects do. I know this means more (temporary) divergence  
> from unstable sometimes, but so be it - if they can't diverge they  
> don't have much purpose.

Sure, this is probably a better way to go -- less confusion for
everybody involved.  Unfortunately, I'm committed to the more confusing
way of working for the 2.1 release, but I will consider the freeze model
for the next release.

If the release manager and stable maintainer roles were to be done by
different people, this would just mean that they would have to
communicate/coordinate, which is healthy anyway.

Living and learning!

-- 
Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
PGP Key ID: 08AC04F9

Attachment: pgprTrou8X5LT.pgp
Description: PGP signature

_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users

Reply via email to