Re: NEW ocaml licence proposal by upstream, will be part of the 3.08.1 release going into sarge.
On Thu, Aug 19, 2004 at 10:02:05PM -0400, Walter Landry wrote:
> Sven Luther <firstname.lastname@example.org> wrote:
> > On Wed, Aug 18, 2004 at 10:02:01AM -0400, Brian Thomas Sniffen wrote:
> > > Sven Luther <email@example.com> writes:
> > > >> I fail to see how requiring modifiers to contribute to proprietary
> > > >> software helps free software.
> > > >
> > > > Because the proprietary version can only happen if the _SAME_
> > > > patch is also applied to the QPLed version.
> > >
> > > To *some* QPL'd version. It doesn't have to be the public one.
> > Well, ok. But the same happens on the GPL, no ? And in general,
> > upstream would not want to maintain a forked version, which is
> > exactly why he chose this licence, so ...
> No, the same thing does not happen with the GPL. Trolltech can take
> contributions under the QPL and include it both in the free X11
> version and the (very expensive) Windows version. If you tried to do
> that with the GPL, you would have to make the Windows version GPL'd as
So ? The BSD allows this too.
> > > > Also, this clause allow upstream to apply the patch to his tree
> > > > without over burdening him to keep two separate trees.
> > >
> > > What's burdening him is his desire to have a proprietary version, not
> > > a contribution of free software for him to use in other free software.
> > And ? Is that so wrong ? Will you declare the BSD non-free too ?
> It is the forced license of modifications that is the problem. BSD
> makes little demands on modified versions.
But the GPL does.
> > > > So this means that more patches can be incorporated, and thus
> > > > the community benefits from it. The fact that there is a
> > > > proprietary version which is _THE SAME_ as the QPLed one, is
> > > > hardly even noticeable, especially in the ocaml case, where i
> > > > doubt there is any significant business going around the
> > > > proprietary version.
> > >
> > > It's not the same. It has extra features -- anything from the free
> > Maybe in the Qt case, but not in the ocaml case.
> Are you saying that if Qt were licensed under the same terms as ocaml,
> Qt would be non-free while ocaml would be free? Even though they are
> under the exact same license?
No, i am saying that you are chasing an imaginary problem in the ocaml case.
> This is yet another example where you are assuming that the initial
> developers are angels. Debian must assume that they are devils, not
No, debian must not, debian-legal has decided to, which is a whole bit
different. And remember, if upstream has deciding to make a gift of their
software to the community, thay can't be fundamentaly evil in the first place,
can they ?
> the least because developers have become devils in the past. I wonder
> how many times we are going to have to repeat this before you get it.
Yeah. If you count things, do you believe that there are more authors of free
software who decided to go proprietary than authors of software with
problematic licences who have resolved their licencing issues and fully freed
their software. If so, i would greatly appreciate that you provide numbers to
support your claims.
And notice that the most spectacular case of upstream going wonk and taking
some software non-free where covered by decidedly free licences, like the
X/MIT licence for example, and nothing is really stopping the FSF to relicence
all the GPLed software under a non-free GLP v3, is it now ? Or even take all
that code we have assigned copyright to it and BSD it, or otherwise take it
Also, many of the software we use is coming from our own community, since some
of us are also upstream authors ? Will you take the insult to the point of
declaring us evil and not to be thrusted ?
> > > software community goes into both, but INRIA's own work or paid work
> > > for paying supporters can go into the private one alone. And if
> > The aim is to provide a version of the code base to the ocaml
> > consortium team members, which has a less restrictive licence, in
> > order for them to do their own in-house variant or whatever. Or
> > simply because their hierarchy is frivolous with open source
> > stuff. All the ocaml-team based development is going into the open
> > source version, except maybe some highly experimental stuff that
> > never got released.
> You're giving me the impression that they want to do something to
> other people's code that they won't allow to their own code. They are
> free to have those goals, but that doesn't make licenses which further
> those goals free.
As long as we can all profit from the software in a free way, what is the
> > > there's no business being helped by this clause, then what's the harm
> > > in removing QPL 3?
> > Well, i would much prefer that they change licence wholy to something more
> > acceptable, that they continue this "our code is under the QPL, but <insert
> > long list of exception>.
> > > > Now, this is much better than the BSD situation, where any code
> > > > can be made proprietary without restrictions, and the BSD is
> > > > free.
> > >
> > > But the BSD license doesn't *force* a proprietary version; it just
> > > allows it. The QPL forces modifiers to grant permission for a
> > > proprietary version.
> > So ?
> > > That force is what makes it non-free.
> > How is your freedom to use the program diminished in any way ? You
> > can do with the software anything you could without the QPL 3
> > clause, and if upstream is incorporating their stuff in a
> > proprietary version, you probably won't even notice. How can you
> > consider something a fee, if you are not in the slightest affected
> > by it ?
> RMS came up with copyleft precisely because of the problems of third
> parties incorporating stuff into proprietary versions. If you don't
> understand the motivations behind copyleft, I'm not sure that there is
> much to talk about.
So what, this is the difference between BSD licences and the GPL. Nothing new
there, both are considered as free.
> > > > And ? Did i not say that the ocaml team was considering moving
> > > > the licence to the little brother of the CECILL family ? And
> > > > that we should postpone the debate right now until those are
> > > > released, and upstream is ready to make the change, probably for
> > > > the next version. I even provided the link and quoted upstream
> > > > on this two times here, but nobody seems to have cared.
> > >
> > > You have not posted a link to a new CECILL-like license that I've
> > Stop being obtuse, and go read the archive. I posted it two times, not a link
> > to this new CECILL licence, but a post of Xavier, where he claimed the
> > possibility of ocaml moving to such a licence, if it was declared non-free by
> > debian and others.
> The post you mentioned was very vague, and just mentioned that they
> would like to move to a DFSG free license in the future. Knowing how
No, their current licence is already DFSG free. But you chose to ignore the
context, so ...
> much people love to deal with legal issues, that future may take a
> long time in coming.
The context behind the creation of the CeCILL licence is a move on the part of
the french governement to move all software created by french governement
agencies or whatever you call them to free software. The ocaml team being done
at INRIA, which is a french governement agency for software, falls under these
considerations. The ocaml upstream have said that they will probably do so, as
long as the licence in questions a) is compatible with their needs, and b) is
declared free by debian and other open/free/whatever software players.
And ocaml 3.09 is mostly due in summer 2005, just in time for our accelerated
sarge+1 release hopefully, so ...
> > > seen. In any case, if this is a plausible solution for you, then why
> > > not move OCaml to non-free for Sarge, then move it back when it's
> > > under a Free license?
> > Don't be ridicoulous, will you. Moving ocaml to non-free for sarge, means
> > moving over 70 packages to non-free that depend on it, including some debian
> > administrative stuff, like i think cdbs, or some of the subversion dependency
> > chain, or coq 8, which was recently freed too.
> According to "apt-cache --recurse" on my testing system, only coq has a
> dependency. Neither cdbs nor subversion do.
I believe in both those cases, it is a build-dependency.
> > It is totally irresponsable to suggest this mere days before the
> > sarge freeze, and only shozs you have no grasp on the realities of
> > debian release management.
> Bugs have to be fixed, no matter when they are found.
If they are real bugs, and many are not.