[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

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 03:28:16AM -0400, Glenn Maynard wrote:
> On Wed, Aug 18, 2004 at 09:29:24PM -0700, Josh Triplett wrote:
> > > I don't see how this makes it non-free.  You are distributing under the
> > > same license you received the software under, so DFSG 3 is satisfied,
> On Thu, Aug 19, 2004 at 04:54:03PM +1000, Matthew Palmer wrote:
> > DFSG #5.  Discrimination against a person or groups of persons.  In this
> > case, the group that contains !(initial developer).  A common definition of
> > discrimination in the sense of exclusion is that exclusion is discrimination
> > when it makes the distinction based on an intrinsic quality, rather than
> > based on a choice.  To take a kroogerism, if we exclude because someone is a
> > White Christian Male, that's discrimination, but if we exclude that same
> > person because they're being an idiot, that's OK.
> The question here: is it free for a license to offer free terms to everyone,
> and extra permissions to another group of people (eg. the original author)?
> Yes.
> If not, for example, this would seem to imply that software under the GPL
> with a special exception releasing John Smith from the requirement to release
> source would fail.
> I think that's clearly silly, because the same effect can be achieved by
> giving that individual a separate license.  Additional licenses being granted
> only to certain people, independently from the one Debian sees, never make
> software less free.  So, it would seem silly to reject a single license
> that combines the effect of this licensing arrangement into one license,
> even if using two licenses would be cleaner.

The effect is different.  For a copyleft, incorporating "group X gets extra
rights" into the licence under which the work is distributed means that any
changes I make have to also be under that discriminatory licence -- that is,
the licence that Debian uses is discriminatory.

Having two separate licence grants to two separate groups is not a problem,
because the licence that Debian receives is not discriminatory.  What other
people get with the work is not our concern.  What the licence forces
everyone to do is a problem.

> Similarly, an effect roughly like the QPL's could be achieved by dual-
> licensing: one license that says "all modifications must be available
> under a permissive license", and another that says "all modifications
> must be available to the original author under a permissive license".
> The first license doesn't fail DFSG#5[1].  Even if the second license
> does, we'd still allow the first, and the second would still be available.

This one's even better, because we can *choose* which licence to use.  One
or the other.  The licence we use (whichever one we want) isn't

> (The first license is a sort of subset of the QPL: you can release your
> changes to everyone and avoid the problem.)

Releasing your changes to everyone doesn't help you avoid the fact that the
original author gets carte blanche to my changes, when nobody else does.

> Rejecting a license because it does this in a single license would be
> strange, if it would be allowed in dual-licensing form.

Not particularly.  With only a slight modification to the QPL, you can get
that effect.  Remove the requirement for the initial developer to re-release
your changes in the QPL version, and tell me that's a free licence.  And yet
everyone gets free rights to the software, some people just get more free
rights.  Because the QPL is a copyleft, that's a real problematic clause,
because I can't licence my changes separately.

- Matt

Attachment: signature.asc
Description: Digital signature

Reply to: