Benjamin S. Kirk writes:
> > BTW, when you are at it, I strongly suggest some decentralized version
> > control system, basically either mercurial or git. In SymPy we are
> > just in the process of moving from svn to Mercurial, you can read why
> > here:
> > http://code.google.com/p/sympy/wiki/Mercurial
> > svn is good, it works well, but mercurial is just better in every
> > aspect (my opinion).
> I don't doubt that... However, the ubiquity of cvs/subversion outweighs
> any drawbacks to centralized control in my mind. I work on enough
> exotic supercomputers that this is a real benefit. Almost any platform
> I go to these days already has cvs/svn installed. It is really a drag
> to have to install your own version control software everywhere you go
> just to access the software you are *really* trying to install.
Besides the compatibility issues, going away from cvs/svn would
probably also mean leaving sourceforge, since those are (AFAIK) the
only two SCM softwares they support. The visibility of libmesh and
the download and statistics services sf provides have been well
worth the cost, in my opinion.
I know there is growing momentum in the area of distributed SCM, and
it is very useful to developers without write access, so it seems it
can't be long before some form of it is supported by sourceforge.