Restriction on distribution by Novell?

[Matthew Flaschen]

Ben Tilly wrote
> I believe it is false that B has to be licensed under the GPL, and it
> is only sometimes true that the patch itself does not have to be
> licensed under the GPL.

You're right.  I was assuming B was based on A.

> 
> Let me take an extreme case.  If I take a file full of source code for
> Microsoft Windows and diff it against a file full of source code for
> Linux, I think it is clear that A is GPLed, B is (hopefully!) not
> GPLed, and the patch falls under both copyrights.  Were I to try to
> distribute that patch, Microsoft could sue me for copyright violation
> and I think they would win. (*)

Right, because MS Windows is not licensed under the GPL (because it is 
not actually derived from A).  In practice, this makes the patch 
essentially meaningless.

> * Yes, I picked an extremely stupid example.  For a more reasonable
> example, consider a case where a company releases a derivative of a
> proprietary program under the GPL.  


The existence of a diff between
> the original proprietary program and the GPLed version
> notwithstanding, the proprietary original remains proprietary.
> 

This is again a very special case, since one entity owns the copyright 
for both programs (and thus can do whatever the hell they want with 
them).  This is the reason a GPL derivative of a proprietary program 
exists here; it is ordinarily illegal.

I still hold that if A is a GPL program, and B is based on A (and their 
are no rights or licenses involved outside of the GPL), then the diff 
from B to A is always under the GPL (unless the diff is trivial in which 
case it may be in the public domain).

Matthew Flaschen