On Sat, Sep 25, 2010 at 11:51 PM, Julia Lawall <[email protected]> wrote:

> Encouraging outside contributions is really just beyond the current goals
> and manpower of the project.  Anyone is welcome to submit a patch against
> the released version.  But until managing that becomes a problem, I would
> prefer to keep things as they are.

I might add that the linux kernel was developed like this for years at
end. People simply sent around patches and patches got applied. The
OpenBSD team has continued this trend (last time I looked, which is
about a couple of years ago). If you do point releases at a certain
rate (depending on the number of contributions), I think this might
actually work out fairly well. The switch to first BitKeeper and then
later on Git seem to be mandated by the fact that at a certain point,
the patches grew to a large number. At that point, RC systems became
the viable option to harness the complexity.

My personal trick would probably be to do what happened in the Erlang
community before they released all code on GitHub.com -- A kind soul
provided a repository tracking all releases of the Erlang system.
People provided patches on top of this repository and there some
healthy development. The fact that you can just export a series of
patches such that it can be applied against the latest released
version makes cross-pollination easy. And it worked surprisingly well.
At some later point, the Erlang people at Ericsson took the full bite
of the apple and put a full repository at github. For their Erlang
system, it was a tremendous success as a fair amount of patches now
come from the outside. However, I do not know the overhead in
maintaining and applying "foreign" patches for them.

For cocci, there is quite a number of papers you probably should have
read before diving into certain parts of the code base. I, for
instance, only know the shallow overview of CTL-vw (and perhaps CTL-vw
is not even used anymore!) and I have only dabbled a bit in the
indentation-preserving parser part of the code. It is far from enough
to be able to hack it with a safe knowledge of not messing up
something. I believe that most non-cosmetic patches would have to put
some toll on Julia, one of her students, another cocci-collaborator
etc., because of the complexity of the code.

-- 
J.
_______________________________________________
Cocci mailing list
[email protected]
http://lists.diku.dk/mailman/listinfo/cocci
(Web access from inside DIKUs LAN only)

Reply via email to