[Biopython-dev] biopython on github

[Peter Cock]

2009/3/17 Tiago Antão <tiagoantao at gmail.com>:
> I can go ahead and try to put a summary of our discussions on that
> page, if nobody opposes. The change can be rewritten afterwards or
> deleted anyway. The only issue is that I can only to that on the
> weekend and not before (travelling abroad from Wednsday to Friday).

Sure - by the weekend I hope we'll have come to a consensus.

> What I think is needed is actually a final decision on how thigs will
> progress. Will there be an official git branch? The official will
> still be cvs? Where will it be hosted? These are lots of important
> questions, but I think there is enough discussion to arrive at a
> decision.

I think it is still to early for a final decision, but here is my
suggested plan:

In the short term (at least until Biopython 1.50 beta is out, perhaps
until Biopython 1.50 proper is out), CVS will remain the official
repository.    Bartek will continue automatically updating the
mirrored copy on github, which will otherwise be treated as READ ONLY.
 If needs be, he may have to reimport the whole history (the tag issue
troubles me - see the other thread), so there may be some bumps along
this road.  Contributions/bug fixes can continue via bugzilla with a
patch, and contributors can also try providing a URL to their own git
branch if they prefer.  During this period I hope most (ideally all)
our active developers with CVS access will create an account on
github, and try out forking from the CVS mirror, creating their own
branches, checking in some changes, and doing some simple merges - for
example pulling code from other Biopython developer's public branches.
 This should give us the confidence to trust git and github enough to
use it for real.

i.e. For the roughly the next month, we will continue as before with
CVS for the real work, but will also try out github.

Once Biopython 1.50 final is out (hopefully by the end of April 2009,
probably sooner), we need to decide if we will actually make the more
to git on github.  At this point, I would expect this to happen by
declaring CVS read only, a static archive (and emergency fall back).
Bartek would turn off his automatic syncing.  We would then continue
working on the github branch with the full CVS history, with a core of
Biopython developers having write access to the "official" branch,
doing new work under their own personal branches for eventual merging
into the main trunk.

I'd still like to have a copy of the "official" git repository running
on biopython.org, but this may not be that easy without some technical
expertise in house to do this.  From initial discussion with the OBF
team about the idea of running git on their servers, my impression is
if we can do it ourselves, we may.  Jason Stajich actually suggested
we use github independently.

Peter

P.S. Could you all update your entry on the wiki participants page
(and if you have one, your wiki user page) to include a link to your
github account:
http://biopython.org/wiki/Participants