Re: [tlug] GPL non-sense

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

Shawn Brown writes:

 > >> What they want to do though is embed their trademark pretty
 > >> deeply.  If they (or anyone) modify the graphics rendering to
 > >> include their trademark, suddenly ...
 > >
 > > Suddenly what?
 > 1) I don't think I can redistribute their binary

That's right.  But then, they can't either as you've described
earlier, as their added clause is in violation of the GPLv3, which
only allows variants that protect trademarks; they can't restrict
distribution of *anything else*.  If you licensed any code to them
under GPL, and they've distributed under the "for personal use"
clause, you can sue them.  (Do you own any of the code or have
contractual interest in its mode of distribution?  That's unclear to
me.)

If they fix their broken clause, then it's not obvious to me that you
can't redistribute their binaries and source *verbatim*, as long as
you acknowledge that those trademarks are property of the relevant
owners in the documentation received with the distribution, and
otherwise don't mention them, especially not in advertising.  That's
not "use" of the trademarks as far as I can see, any more than giving
somebody a can of Coke is "use" of the "Coca-Cola" trademark displayed
on the outside.  (Evidently the Mozilla Foundation disagrees, but
their license is also not the GPLv3 -- AFAIK it explicitly requires
removal of their trademarks in case of binary distribution, rather
than denying permission to "use" them.)

However, as soon as you make *any* changes, I'm afraid things are
really unclear, because the fact is that you've made these changes but
have not removed the trademarks deliberately (even if that's very
understandable :-).  Eg, if you introduce a bug, and somebody loses
data thereby, you reduce the value of the trademarks (guilt by
association *is* allowed in the "court of consumer opinion", you see).

 > 2) I don't think I can redistribute their src

See above.

 > 3) to make a distributable binary and src, I will have to figure
 >    out where and how they are rendering their trademarks and
 >    replace it with something else.

Yes, in practice; verbatim redistribution isn't very useful in
practice, so the above discussion is quite theoretical.  But while
this is a PITA, it is not "hard."  It should be easy to identify the
trademarks, and if it isn't because the trademarks and/or the
executable code is obfuscated, then it doesn't count as "Corresponding
Source" (at least in the FSF interpretation of the GPL).

 > This is despite the GPL clause:
 > 
 > TERMS AND CONDITIONS 1. Source Code.  The "Corresponding Source"
 > for a work in object code form means all the source code needed to
 > generate, install, and (for an executable work) run the object code
 > and to modify the work, including scripts to control those
 > activities.
 > 
 > Their source code and scripts does not produce a distributable
 > binary or source.  Isn't that the intent of the requirement.  No?

No, that isn't the intent of that requirement.  The intent of the
requirement is that you be allowed to study, modify, and redistribute
the source code of the program, to understand how it works, to fix
problems and improve it, and to provide the benefit of your acquired
knowledge to anybody you like in a form of your own choosing.

It doesn't say that they have to make this easy, it says they have to
make it *possible* and *permitted*.  If it did say they have to make
it easy, most GNU code would have to be withdrawn from distribution
for lack of intelligibility. ;-)

 > Like I said, I think the requirement should be modified to specify
 > that the resulting source and binaries derived from gpl3 code be
 > redistributable.
 > 
 > Before this, I thought it was.

I hate to tell you this, but this is a *deliberate feature*, not a
bug, in GPLv3.  The optional clauses restricting use of embedded
trademarks were not permitted in earlier versions of the license.  If
you embedded trademarks in the program and distributed that program,
you licensed everyone to make copies and derivatives without
limitation (arguably including a derivative that contained only the
trademarked logo, although I'm not sure that a court would agree).

In order to legally protect your trademarks and still use them in a
GPLv2 program, you needed to embed them in loadable data, and either
make the program work in the absence of that data or provide
substitute data for use in the redistributable portion of the program.
Then the trademarks (because they are data, not part of the executable
program as such) could be provided as (proprietary) addons, which
should satisfy everybody.

Why the FSF caved in on this reasonable requirement (and good
programming practice!) I don't know.  Why the corporate participants
in the GPLv3 revision discussion wanted the GPLv3 version is obvious:
they wanted to be able to do to you what they're doing to you.  (I
imagine the ostensible rationale was that the GPLv2 way of doing
business was unreasonably costly and didn't have any perceivable
benefits to users or enhance software freedom.  Bullshit, of course.)