Compatibility of the AFL with the GPL

[Eben Moglen]

On Friday, 14 March 2003, Russell Nelson wrote:

  Lawrence E. Rosen writes:
   > OK, guys, play with me one more round.  This time, let's do it in the
   > form of a law school exam question and let's get the lawyers and IANALs
   > on this list to chime in:
  
  Nahhh.  None of this is necessary.  There's nothing in the AFL that
  says that you must use the same license on derivative works.
  Therefore, without reference to any other terms of the AFL, it is
  trivially compatible with the GPL insofar as derivative works get
  licensed under the GPL.
  
  Period.
  
  End of discussion.
  
  It's GPL-compatible.
  
  The only question is whether RMS will admit to making a mistake.
  
No, that's not quite right.  We do have to resolve one question, which
is whether the effect of the AFL is to pass through the patent-
retaliation provision on code which is relicensed.  If so, the AFL
does not permit relicensing under GPL, because GPL 2(b) requires that
no additional terms be added to GPL'd code during modification and
redistribution.  So, here's the one case we need to be sure we agree
about: A publishes program FOO under AFL.  B modifies FOO by combining
FOO with preexisting GPL'd code to make GOO, and releases under GPL.
C modifies to make HOO, distributes under GPL, and later sues A
alleging patent infringement by A in another program released under
AFL.  What's the result here?  If C cannot continue to distribute HOO,
the AFL has succeeded in imposing additional conditions not only on
the FOO part of HOO, but also on other GPL'd code with which it is
combined; this would be a modality for evading GPL.

E

--
license-discuss archive is at http://crynwr.com/cgi-bin/ezmlm-cgi?3