Re: XFree86 license difficulties
On Sat, Jan 31, 2004 at 05:17:00PM +0000, Andrew Suffield wrote:
> On Fri, Jan 30, 2004 at 12:31:44PM -0700, paul cannon wrote:
> > Several posters on slashdot and elsewhere have mentioned the
> > similarity between this and the old, obnoxious BSD "advertising
> > clause":
>
> Let's nip that in the bud before people start wasting time over
> it. It's not really the advertising clause. It happens to be a problem
> for more or less the same reasons, but that's true for most of the
> GPL-compatibility problems we come across.
I was trying to state this, in my own clumsy way. I am well aware it is
not the same as the advertising clause.
> > As we all know, the FSF [6] considers the mere act of linking to
> > create a derived work for the purposes of the GPL, and claims
> > anything linked to a GPL'd work must also be distributable under the
> > terms of the GPL.
> >
> > If the XFree86 Project takes a similar stance (which, indeed, does
> > not seem to be the case right now) then anything linked to an
> > XFree86 library must be distributable under the terms of the XFree86
> > license. That case would add somewhat deeper problems than simple
> > license incompatibility; it would mean no program could link against
> > both Xlib and a GPL'd library. This would seem to make it impossible
> > to distribute Qt, for example.
> >
> > If XFree86 does not consider linking to create a derived work which
> > must carry the same restrictions as those in the library, then it
> > does not seem there is a problem; an application linking against Qt
> > and Xlib could be solely under the GPL. Or am I off my rocker here?
> >
> > Is it likely that the XFree86 Project will take that stance on
> > linking?
>
> The opinion of the XFree86 project is irrelevant. It is the licenses
> on GPLed works that would be violated, not the license on XFree86, so
> it's the interpretation of the authors of the GPLed works that counts.
I don't quite see how this is so. If the XFree86 Project were to say-
theoretically- something like "linking dynamically to an XFree86 library
does not constitute a derived work for the purposes of the XFree86
license" then Qt (e.g.) could be dynamically linked to Xlib and be
legally distributed under the terms of the GPL. The GPL is satisfied,
since the linked product could be distributed under the exact terms of
the GPL, and the XFree86 license is also satisfied, since they wouldn't
claim any copyright over the linked product.
Is there a legal precedent or doctrine specifically stating that linking
dynamically against a library produces a derived work? I have seen that
the FSF claims this is so, and so it makes sense to apply their rule to
products linked with GPL'd works, but it doesn't make sense to me that
such thinking should be universally applied (the library is only being
used or accessed, not necessarily being modified or distributed).
--
paul cannon
pik@debian.org
Reply to: