[cc-community] GNU-GPL, makefiles, and non-distributed executables
hancock at anansispaceworks.com
Mon Dec 17 13:26:50 EST 2007
rob at robmyers.org wrote:
> Quoting Terry Hancock <hancock at anansispaceworks.com>:
>>Eve's code is merely "designed to work with" George's work. However,
>>that is "use" and is not covered.
> Eve's work is object files that are designed to be useful only when
> linked against George's work. This is not working with, this is
> combining with and creating a derivative of. It is an issue of
> copyright not functionality or interaction.
Usefulness has nothing to do with inclusion or derivation for copyright
purposes. Obviously, the critical role of scientific citations is not
clear to all readers here. Perhaps the following examples will be more
* A "concordance" can be made of another work (a Bible concordance,
though very common, is a bad example because many versions of the Bible
are public domain anyway, but you can find concordances of Tolkien's
work, Star Trek episodes, etc. Wikipedia has episode-by-episode
descriptions of many different TV series. No one argues that these
"include" the works they describe -- yet they are just as useless
without the original work as a library API without a library).
* Consider a set of usage instructions for a VCR. Without the VCR
itself, the instructions have no use or reason for being at all. But the
instructions are not "derived" from the VCR in the copyright law sense
(nor from its design, which *is* copyrightable, even though the machine
itself obviously isn't).
* A critical review of a film has no utility without the film, but it is
still not a derivative.
Hence, a necessity for functional utility is NOT a criterion for
considering a work to be "included" in another work.
> GNU argue that even RPC and pipes create a derivative if the program
> has "carnal knowledge" of the internals of the library. This isn't
> unlimited, named pipes with dumb command passing is OK for example.
Perhaps, but that's not what they wrote into the license (and I think it
can be strongly argued based on other GNU statements that they did not
intend to include it either -- the argument that *using* a copyleft work
doesn't invoke any copyleft has been made strongly and loudly in defense
of the principle of copyleft as a "free" provision). What seems most
likely is that they noticed that this might create a problem for them
*after the fact* and thus *changed* their intent.
What GPLv2 says is:
a "work based on the Program"
means either the Program or any derivative work under copyright law
And only works "based on the Program" are covered by the GPL. Consider
If identifiable sections of that work are not derived from the Program,
and can be reasonably considered independent and separate works in
themselves, then this License, and its terms, do not apply to those
sections when you distribute them as separate works.
In GPLv2, at least, we are strongly encouraged to believe that
"reasonably considered independent and separate works" should be judged
by the standards of copyright law (probably US law?).
GPLv3 tries to redefine these concepts to express better what GNU
considers to be "reasonable". IMHO, their new provisions are legally
binding. Whether they are "reasonable" or "free" is something I'm not as
sure about -- I don't think that will be clear until we see how the
license is used.
>>"Referencing" a work is NOT "including" it (imagine the consequences if
>>it were!). Lists of bookmarks are a trivial example of why that would be
>>bad as are scientific citations, bibliographies, etc.
> Linking creates a derivative work and this is an act covered by copyright.
> Linking actually produces a combined "text" containing the whole of
> the work interleaved. This is different from bibliographic references.
Yes and yes it does. But you can link a GPL work with any other thing
you please as long as you don't distribute ("use" is not restricted).
GNU also argues this, quite strongly, when defending the "freedom" of
the GPL's copyleft.
> GNU have historically drawn the line somewhere before distributing
> object files to be linked against GPL-covered software as the NeXT
> example shows.
I agree that GNU would like to restrict this case. I think that quite
possibly they have succeeded in doing so in the GPLv3. But "GPL" still
means "GPLv2" to me, and I don't think it covers this case.
Terry Hancock (hancock at AnansiSpaceworks.com)
Anansi Spaceworks http://www.AnansiSpaceworks.com
More information about the cc-community