Re: NEW ocaml licence proposal by upstream, will be part of the 3.08.1 release going into sarge.

[Brian Thomas Sniffen <bts@alum.mit.edu>]

Raul Miller <moth@debian.org> writes:

>> that's not Free.  The QPL says "BSD to inria/cristal, copyleft/patch
>> to everyone else, and that pair must be passed along" -- that "must",
>> that added restriction, is the non-free part.
>
> I'm not sure about this.
>
> As stated, that "must" is inherent in copyright law -- if we start saying
> that "free software" requires that non-copyright holders be allowed to
> change license terms, I think we've missed the boat.

That "must" is not inherent in copyright law.  This is talking about
derived works -- it would have been more clear if I'd said "passed
along with modifications" in every example.  So I own a copyright
interest in the work as well, and copyright law certainly does not
require that I grant the initial developer a license to exploit my
work.

So I'm not insisting that non-copyright holders be allowed to change
terms; only that modifiers be able to release their code under the
same license that they had to the initial work -- you know, what DFSG
3 says.

> Can you express this point focussing on issues of real, practical
> non-freeness?  [for example, stuff which interferes with porting, bug
> fixes, security fixes, ...]

There were demands that I point at the DFSG, and I did so.  Now you
want a practical example.  Here's some attempts at one:

* I cannot exercise my free-software rights (modify and distribute the
  code) *and* restrict what the initial developer does in any way.

* I can't fork the code, even distributing as patches.  There's no way
  for me to make XEmacs, which is FSF Emacs + code by people who won't
  transfer copyright to the FSF.

* This is the case that bothered me: I can't distribute a modification
  which incorporates a change to which I can't give INRIA a permissive
  license.  In fact, I can't merge in *any code whatsoever* that I
  didn't write myself, because I have to be able to give INRIA that particular
  permissive license to it.

  For example, the Debian O'Caml maintainer can't integrate any patch
  submitted, because he can't grant INRIA the license they demand.

> What you're essentially saying, from my point of view, is that where we
> have two licenses, both of which are free, one discriminates against
> users and the other does not.  But if that were the case, one of them
> wouldn't be free.

I don't understand you here.  I don't think you're characterizing what
I've said correctly; is there some bit I can explain in more depth?

-Brian

-- 
Brian Sniffen                                       bts@alum.mit.edu