[ocaml-infra] ocaml.org licensing

[Anil Madhavapeddy]

On 1 Mar 2014, at 20:29, Fabrice Le Fessant <Fabrice.Le_fessant at inria.fr> wrote:

> For Python, following your example, there is a http://www.afpy.org/
> for the french community, with its own forums and mailing-lists. I
> imagine that there are similar websites for most non-english
> countries. For Linux, there are hundred of websites, but only one
> community, isn't it ?

The http://afpy.org example you show doesn't appear to share any
content with the http://python.org website.  So how is that a fork?

"ocaml.org" consists of a few different pieces:

1) tools
2) content
3) compute infrastructure

1) and 2) are pretty easy to re-use as anyone wants once this licensing
thread is closed.
3) will be quite difficult for other people to just fork, since there's
real resources and effort that goes into keeping it running.

As I see it, OCaml's had a very successful 16 years of distributing
community content all over the Internet.  My goal for ocaml.org is to
see it become a single point of focus that serves OCaml users with
accurate, up-to-date documentation, tutorials and packaging infrastructure.

To date, this focus has been going very well -- in recent months, we've
put the new design live, moved the OPAM infrastructure over to the
subdomain, activated SSL certificates, deployed Rackspace infrastructure
thanks to their generous donations, and got a great Travis testing
workflow going for ongoing testing.

Work has been ongoing to improve opam2web (via opamfu) to further integrate
OPAM package and documentation into the main website, as well as provide a
high-quality French translation as the first major translation. So given
all this great work, we should remain focussed on hitting this goal
and continuing to grow our userbase.

I'm trying to understand what the concrete objection here is.  Is it
to do with the licensing choice for 2)? Bear in mind that forking
infrastructure will always be difficult -- despite the existence of
afpy.org, there's still a far more impressive set of lists at the central
Python mail server: https://mail.python.org/mailman/listinfo

So to my question: is there some group that you feel we aren't serving
adequately at the moment that would benefit from a refinement of our
approach?  It's important to know this before we lock in a decision on
licensing, as that's a difficult decision to reverse.

-anil