Virally licensed code in FreeBSD kernel

[Chad Perrin]

On Sun, Apr 15, 2007 at 03:27:34PM -0700, Ted Mittelstaedt wrote:
> >
> > Between these four sections -- 1.3, 1.6, 1.9, and 3.6 -- it is not 100%
> > clear what the intention (in a legal context) of the licensing as it
> > applies to a "Larger Work" must be.  Granted, it sure looks like no
> > other parts of a "Larger Work" that are not part of the same file as
> > code explicitly licensed CDDL would be covered by the CDDL, but a lawyer
> > could probably make a pretty good case for the license extending beyond
> > the explicitly licensed code in a case where a "Larger Work" is compiled
> > as a single binary executable.
> >
> 
> Chad, that is very unlikely.  If that kind of tack would have worked then
> SCO would have tried this when they filed the infringement lawsuit against
> Linux.

How would SCO have leveraged the phrasing of the CDDL in its lawsuits?
Last I checked, none of the SCO Unix code had been released under terms
of the CDDL.


> 
> This argument was also NOT tried when Unix Systems Labs sued University of
> California, Berkely over the original FreeBSD source code.

See above, re the CDDL, and let SCO = Unix Systems Labs in this example.


> 
> In both those lawsuits, the lawyers didn't attempt to include the -entire-
> distribution.  Instead they included the subsections.  In the USL lawsuit it
> was the kernel.

It's not the simple fact of code being used -- it's the fact of the
word choice in the CDDL being somewhat ambiguous.


> 
> The SCO lawsuit is still languishing but the USL lawsuit was settled after
> the judge ruled that USL couldn't make a case, and
> subsequent to this the lawyers set out a list of specific source files that
> infringed.

Yes, I'm aware of all that.  None of it is relevant to a discussion of
the ambiguity of the CDDL, however.  In fact, the CDDL didn't even exist
at the time of the USL lawsuit.


> 
> > In any case, even if the rest of the code in the compiled binary is not
> > in fact covered by the CDDL under any circumstances other than explicit
> > release under terms of the CDDL, one still has a source distribution
> > obligation for part of such a compiled binary according to the law and
> > the terms of the CDDL -- the part that is distributed under terms of the
> > CDDL, at minimum.  This means that a compiled binary that includes CDDL
> > code in its source files carries a source distribution legal obligation,
> > period.  We're chipping away at the "freeness" of the software either
> > way.
> >
> 
> Here is the summary problem, and it's the problem with the GPL.  Copyright
> law basically says the owner cannot release copyright interest by doing
> nothing.  In other words if I compile GPL code into my code and distribute
> the result, my code's copyright stays with me, unless I do something
> explicit like
> signing over copyright ownership to the FSF

This has nothing to do with any "release" of copyright interest.  It's
about distribution licensing terms.


> 
> With GPL code where the owner has given over copyright to the FSF,
> the FSF can simply threaten to sue an infringer if they distribute a
> modification
> to GPL code then try to use their copyright rights over someone else to
> prevent
> them further modifying their modification in ways they do not want to
> have happen - because the FSF can withdraw
> permission they have to use the FSF-owned copyrighted code under GPL.

You don't have to give copyright to the FSF -- all you have to do is:

  A) release something under terms of the GPL (while keeping the
  copyright for yourself, even)

  B) include your code with GPLed code in such a way that the terms of
  the GPL apply to your code (while keeping the copyright for yourself,
  even)

The fact that the FSF would like everyone in the world to assign
copyright for every piece of software they create to the FSF so that the
FSF can enforce the terms of the GPL more effectively in no way implies
that you have to assign copyright to the FSF for the GPL to apply.

It's point B that applies, by way of analogy (though not directly), to
the case of whether the CDDL would apply to the case we've been
discussing here.


> 
> But in the vast majority of cases of GPL code the developer retains
> copyright
> on the stuff that is distributed via GPL.  It is not at all clear that if
> someone takes
> that GPL code and does something in violation of the GPL, and the copyright
> holder refuses to do anything about it, that the FSF has any legal standing
> to
> get involved in a lawsuit.  And if they don't, and nobody else sues, then
> the
> GPL violator will "get away" with violating the GPL.  And if that happens
> enough times, the "threat value"of the GPL will become mostly useless.

Not so.  The "threat value" of the GPL is exactly as "useful" as
appropriate to the willingness and ability of a copyright owner to
enforce the terms of the GPL in each case.  If five billion people write
software and release it under the terms of the GPL without assigning
copyright to the FSF, then never bother to enforce it at all, and the
other billion of us write software and assign copyright to the FSF, the
terms of the GPL for those last billion cases would carry just as much
"threat value" as though none of the other five billion people had acted
as I described.


> 
> Right now the only "threat value" that the GPL carries against infringement
> for non-FSF code is that people know that even if the original copyright
> holder
> didn't sign over copyright rights to the FSF, and doesen't have deep
> pockets,
> if they infringe GPL code and the copyright holder
> doesen't like it, the FSF will come rushing in with their money and lawyers
> to
> help the copyright holder to sue the infringer.

Tell that to Sun Microsystems, releasing the JVM under terms of GPLv3.
I'm pretty sure they don't need the help of the FSF to enforce the terms
of the license.


> 
> What people don't know about is what is happening with GPLed code that
> the copyright stayed with the owner, was never given over to the FSF, and
> the original author loses interest in it, and someone comes along and
> infringes it.
> 
> There's dozens if not hundreds of GPL projects languishing around up on
> Sourceforge that haven't had updates to them for years, have had few
> downloads, and little general interest by the community and no interest by
> their copyright holders.  If someone came along and stole sections of
> the work and put it in a commercial product, do you think the FSF would
> sue them, or even publicize it?  I very much doubt it.  The FSF does not
> like to advertise failures in the GPL and I believe that there has been a
> lot of infringement of GPL code in the past that the FSF has remained
> quiet about - simply because they cannot interest the copyright holders in
> getting involved in a lawsuit, or transferring copyright ownership to the
> FSF.

I'm really not sure what all this has to do with the discussion at hand,
frankly.


> 
> A software license is only as good as the willingness of the copyright
> holder
> to back it up.  For FSF copyrighted code, there's a lot of willingness - but
> the general public often mistakenly equates
> fsf-copyright-owned-code-under-GPL
> in the same league as jow-blow-owned-copyright-code-under-GPL.

This is true.  I'm still not sure what this has to do with the
discussion at hand -- namely, what terms the CDDL imposes under various
circumstances, particularly circumstances related to ZFS support in
FreeBSD.

-- 
CCD CopyWrite Chad Perrin [ http://ccd.apotheon.org ]
Brian K. Reid: "In computer science, we stand on each other's feet."