[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: GPL and command-line libraries

Anthony DeRobertis <anthony@derobert.net> writes:

> Brian Thomas Sniffen wrote:
>>>But in the case of the DFSG and the GPL it does. Saying "You may not
>>>distribute this work along with a frame designed to hold it" violates
>>>DFSG 1.
>> But saying "You may only distribute this work with a frame designed
>> to
>> hold it if that frame is freely distributed" is Free.
> No it isn't. DFSG 9 ("For example, the license must not insist that
> all other programs distributed on the same medium must be free
> software.")

This isn't talking about all other programs on the same medium.  It is
only talking about those programs integrated with the GPL'd program.

>>>And the GPL does not attempt to cover the frame; the GPL explicitly
>>>defines a "work based on the Program" as "either the Program or any
>>>derivative work under copyright law."
>> And it further clarifies *that* to be "a work containing the Program
>> or a portion of it, either verbatim or with modifications".  So if you
>> ld or tar some programs together, you now have a work containing the
>> Program verbatim.
> The full sentense is:
>      The "Program", below, refers to any such program or work, and a
>      "work based on the Program" means either the Program or any
>      derivative work under copyright law: that is to say, a work
>      containing the Program or a portion of it, either verbatim or with
>      modifications and/or translated into another language. (Hereinafter,
>      translation is included without limitation in the term
>      "modification".)
> I think "that is to say" introduces a laymans explanation of what a derivative work is. In order to accept the alternate explanation --- 
> that it seeks to define derivative work --- we'd have to (a) disregard
> its explicit definition as a "derivative work under copyright law" and
> (b) grant the FSF powers generally reserved to congress and the courts.

I don't think the problem of interpretation is nearly so bad as you
imply.  It seems clear that they mean both 17USC Derivative Works and
other works derivative of the Program.  Since the GPL is intended to
have reasonable functionality in many copyright regimes, I think any
attempt to force it to a single verbatim definition of copyright law
is unlikely to succeed.

So we do not need to grant the FSF exceptional power to redefine law,
and neither do we need to disregard any explicit definitions.  As with
any licensor who gives close but not exactly synonymous definitions,
we treat the definition as the union or intersection of those
definitions, as appropriate.  The safe choice here is the union: so
the GPL covers all derivative works under your local copyright regime,
and also restricts any work containing the Program or a portion of it,
with or without modifications.

It avoids violating GPL 9 by later removing all restrictions on mere

> Further, we can not say that tar can create a work containing the
> program because copyright law concerns itself only with creative
> works; tar can not do such a thing.  Instead, we just have two
> seperate works.

Certainly tar creates a work containing the program.  The user who
chooses which programs to tar together creates an anthology, a
collective work.

Additionally, some works have creative content and some do not.  A
work made without creative content -- say, by just tarring together to
randomly selected files -- does not evade copyright restrictions on
those files.  It is a copy of them.  If you run tar cf /tmp/foo
/usr/bin/emacs21, then you have made a copy of GNU Emacs.  You may
only do this as permitted by its license.

>> Some works are neither derivative works nor mere aggregation; they
>> might be functional combinations, for example.
> Yes, but it's a copyright licence; under copyright law, has anything
> other than mere aggregation happened?

Mere aggregation is not a phrase from copyright law.  It is from the
GPL.  With relevance to copyright law, copying and distribution have
each happened.  The GPL permits such copying and distribution, under
the conditions that the entire modified work be distributed under the
terms of the GPL.

> Which clause of the GPL would I be violating by distributing such a
> "functional combination" with one part GPL'd and the other part not?
> And why isn't Debian violating this same clause?

You're not distributing a verbatim copy, so you can't distribute under
clause 1.  You have to use clause 2.

Well, under clause 2, you have modified your copy of the Program,
forming a "work based on the Program".  You have then copied and
distributed that work.  Under 2b, you must cause the work you
distribute, which in part contains the Program, to be licensed under
the GPL.

Following the text after 2c: "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.  But when you distribute the same sections as part of
a whole which is a work based on the Program, the distribution of the
whole must be on the terms of this License, whose permissions for
other licensees extend to the entire whole, and thus to each and every
part regardless of who wrote it."

Then it talks about control of "collective works based on the
Program", explicitly endorsing the anthology analogy I used.

Then it talks about mere aggregation -- but only with works *not*
based on the Program, and only about aggregation on a volume of
storage or distribution medium.  Aggregating together, say, readline
and a program written to use it fits neither of these conditions.  The
program written to use readline is based on it -- they could have used
the word derivative here if they'd meant to restrict it to 17USC
Derivative Works.  Also, while tar's output could reasonably be called
a distribution medium, I could never call ld's output a distribution
medium with a straight face.

>> Or they might be
>> anthologies, in which creative effort has been expended in the
>> selection of works.
> Creating an anthology isn't listed as one of the copyright holder's
> exclusive rights; presumably, if he grants you the right to reproduce
> and distribute, you may create an anthology. (Remember, the copyright
> over an anthology only applies to the anthology itself, not to the
> individual works)

Yes, and those rights are conditionally granted by the GPL.  They
allow some, but not all, anthologies.  In particular, anthologies
("collective works") may only be distributed under the GPL.

>> If I tar up Emacs and a bunch of its elisp files, certainly that's
>> not mere aggregation.
> http://lists.debian.org/debian-legal/2002/11/msg00217.html contains a
> quoted message from RMS on a similar subject.

Yup.  And he seems to agree: a bunch of "its elisp files", using the
Emacs interfaces, are a work based on Emacs, whether or not they are


Brian Sniffen                                       bts@alum.mit.ed

Reply to: