Dear Sage-Combinat developers,
For about five years, our workflow has been based on mercurial queues.
It basically did the job, but there has been recurrent griping about
it and for good reasons! We all agree that the main issue of our
workflow is that, at this point in time [1], mercurial queues handle
rebasing naively. The two other issues are that it's too complicated
and monolithic. To some extent those issues are killing us.
So, it's great that people are exploring alternative workflows.
Actually we may not even have a choice: there is an ongoing project to
change the Sage workflow in depth, whose purpose is to make
contributing to Sage a much more seamless process. The leaders of that
project decided that this would be achieved by switching Sage's
version control system to GIT. At this point, I am not keen on that
choice because I see a big cost; but in our world those who act get to
make the decisions they see fit, and that's how it should be.
Back to Sage-Combinat. Many suggestions have been thrown in the air
over the years: why don't you use branches? git? ...?. I may have
looked reluctant, but that's just because spelling out good looking
ideas is not the end of the story. Hopefully there are great
opportunities ahead for a much smoother workflow. But we could also
get stuck in a workflow that kills us even more. Thus, before jumping
onto another workflow we need to be really sure that:
(a) It will scale
(b) It will still resolve reasonably all our constraints [2]:
(c) It will not put off contributers more than the current situation
The other thing is that we need someone taking the lead for the change
of workflow. This includes:
(1) Running some large scale experiment to check for (a), (b), and (c).
(2) Participating to the upcoming Sage-GIT workshop.
(3) Writing support scripts.
(4) Training our contributers, and convincing them it's worth it.
(5) Taking responsibility if things go wrong!
and
(0) Linking this thread from [2] :-)
Volunteers?
I already signed up for (2) to get things going. But honestly I
already have trouble handling all that's on my plate.
Cheers!
Nicolas
[1] Mercurial devs, including Pierre-Yves in CC, are working hard on
it, with the emerging concept of mutable history as a basis for
replacing mercurial queues:
http://mercurial.selenic.com/wiki/ChangesetsObsolescence
http://hg-lab.logilab.org/doc/mutable-history/html/
[2] http://wiki.sagemath.org/combinat/CodeSharingWorkflow
Note that the stringiest constraint we faced so far (namely that
it was impossible to assume that all Sage-Combinat dev would be
working with the same version of Sage) may vanish with the new
Sage workflow. Time will tell.
--
Nicolas M. Thiéry "Isil" <nthi...@users.sf.net>
http://Nicolas.Thiery.name/
--
You received this message because you are subscribed to the Google Groups
"sage-combinat-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-combinat-devel+unsubscr...@googlegroups.com.
To post to this group, send email to sage-combinat-devel@googlegroups.com.
Visit this group at http://groups.google.com/group/sage-combinat-devel?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
