[tlug] Proprietary derivatives of FLOSS and other absurdities [was: Why Hollywood does break foreign films ?]

["Stephen J. Turnbull" <stephen@example.com>]

Benjamin Tayehanpour writes:

 > On 13 August 2013 15:12, Stephen J. Turnbull <stephen@example.com> wrote:
 > > Both can inherit from a permissively licensed code base (and do).

 > Yes, obviously, but that's still a one-way thing. I haven't explicitly
 > shared code with you just because we both happen to have code in us
 > from another project. There's no sentimental bond between us, there's
 > no love lost.

True, but in the original context common inheritance was *the* point.
If social context (including common code bases) doesn't matter to us,
why are we doing this?  (That's not a rhetorical question.)

 > On 13 August 2013 15:12, Stephen J. Turnbull <stephen@example.com> wrote:
 > > You can, if you're the copyright holder.

 > No, you can't. At most, you can dual-license the code, in case you
 > happen to have a customer whose use-case is backwards enough not to be
 > GPL friendly, but you can't revoke the GPL. It's a one-way street,
 > baby~
 > http://www.groklaw.net/article.php?story=2006062204552163

Nope.  That article is based on "an attorney's opinion".  That's not
law; in the U.S. law is made by legislation and courts, and there's no
law that explicitly allows licensees to enforce promises made by
licensors.  Nobody has ever tried to actually revoke the GPL.  We
won't know for sure until that happens whether it can be revoked.

Remember, the GPL itself declares that it is *not* a contract.
Therefore it is a gift, and can in principle be revoked.  Without a
contract, users have no standing in U.S. court, the copyright holder
holds all the cards.  The arguments for why it can't be revoked are
based on the concept of contract via "reliance"[1], so you need to
lose something important to be able to defend against a suit.  That
will work for Linux, but most open source code is important only to
the author.  I can think of reasons involving future large profits to
the copyright holder and losses to the FLOSS community for doing so in
that case, and there would be no reliance.  Your guess is as good as
mine -- but no better -- as to what happens in that case.

However, once again, the main point is common heritage.  You want to
know fear and loathing?  Here's a loophole acknowledged by the FSF
legal staff: if the FSF were to disband without transferring
copyright, all FSF-help copyrights would fall into the public domain.
After that, anybody could make proprietary software out of GPLed
software, without even changing a line, because there would be nobody
with standing to enforce the GPL.  There's another one: the assignment
I made to the FSF does not obligate them to prosecute violations of
the GPL.  If the FSF's stock of copyrights were to fall into the hands
of somebody (say, Eric A. Young, look it up) who credibly could
promise to not prosecute, well, ....  Fortunately, RMS is a
monomaniac, his likely successors are also true believers, and his
legal staff is good (they will ensure there will be successors to the
FSF) so I'm not worried about any of my assigned software for my
lifetime.

Now, to the point.  The purpose of copyleft is to ensure that free
code remains free code, even after somebody else incorporates it in
another program.  In existing copyleft licenses and in the FSD and
OSD, there is no obligation to maintain a line-by-line record[2], nor
to provide access to repos[3], so by mixing the code with proprietary
code the copyright holder can effectively make that instance of the
code proprietary.  Or, as the (non-) permissions notice in the Xlib
Programming Manual says

  Inasmuch as the proprietary revisions cannot be separated from the
  freely copyable MIT source material, the net result is that the
  copying of this document is not allowed.  Sorry for the doublespeak!

You may not care if your copyleft material were used in proprietary
derivatives, but many people who use copyleft licenses by choice would.

Footnotes: 
[1]  Mentioned obliquely in the article you link.

[2]  This is now feasible with $VCS blame.

[3]  Only the source for the distributed software.  So use of $VCS
blame is infeasible to third parties without a court order.