Re: GPL traitor !

[Hyman Rosen]

Alan Mackenzie wrote:
> I disagree.  It says "a set" of instructions which gets the certain
> result.  That implies that it includes the other programs and all the
> bits of the operating system and BIOS it uses, otherwise that set
> couldn't bring about any result, no matter how uncertain.  Which is kind
> of ridiculous.

If you were correct, that would imply that no utility library
could be copyrighted. That's what's ridiculous.

It says that a computer program is a set of instructions because
that's a reasonable way of describing what a computer program is
using conventional English.

> Well now the problem is that an arbitrarily small fragment of the whole
> program, even just a single machine instruction, counts as "a set" of
> instructions, hence is, legally, "a program".  This is even more ridiculous.

No, because computer programs are simply one kind of thing that may
be copyrighted, and those things fall under the overall rule
    <http://www.copyright.gov/title17/circ92.pdf>
    § 102 · Subject matter of copyright: In general
    (a) Copyright protection subsists, in accordance with this
    title, in original works of authorship fixed in any tangible
    medium of expression

The computer program must be an original work of authorship. The
tinier the program, the less likely it is to be so, in the same
way that a tiny sentence fragment would not be copyrightable.

Moreover, as described by the SFLC,
<http://www.softwarefreedom.org/resources/2007/originality-requirements.html>
the abstraction-filtration-comparison test will remove from
consideration of copyright violation any elements which are
dictated by efficiency or external factors, further limiting
the possibility of copyrighting very small programs.

Aside from that, however, it is no more ridiculous to copyright
a tiny computer program than it is to copyright a tiny poem; do
you believe that a Haiku cannot be copyrighted?

> I can't accept that.  Programs are frequently several files, a main
> program together with dynamic libraries, which form an integrated whole,
> i.e. each of the files is non-functional except when linked with all the
> others.

If someone ships such a set of files in tandem, a plausible argument
could be made that this is a collected work such that the entire
collection must be distributed under the GPL if any part of it is
covered by the GPL. Shipping only the non-GPLed parts, however, does
not require permission from the copyright holders of the GPLed dynamic
libraries with which the program will interoperate when executed,
because the GPLed portions are not being copied.

> OK, I can accept that.  Now what is the equivalent legal term for
> "complete program"?  I put it to you that it's "program".

There is no such term because it is not necessary; there is no
distinction in copyright between them.

> Surely not.  A complete program, a smallish one, could easily just be a
> single source file written by a single hacker over a weekend.  Or are you
> going to tell me that the "number of contributions" can be one?  That's
> absurd too, because then every work would be, in legal terms, a
> collective one, rendering the term "collective work" equivalent to
> "work", and hence redundant.

What in the world are you talking about? A program embodied by a single
source file, or single binary file, for that matter, is not a collective
work. It has a single copyright held by its author. When the process of
translation of a program to binary involves the copying of other programs
into the resulting file in combination with the work being translated,
the result is a collective work. The copyright over the collective work
is held by the author who created it, but the copyrights over the pieces
belong to the copyright holders of those additional programs. In order
for this collective work to be copied, permission is required from all
of the authors. It is here that the GPL takes effect, disallowing the
copying of the collective work unless the work as a whole is distributed
under the GPL.

> The essential thing about a complete program is that, of itself, it
> works when loaded onto computer, regardless of the nature of
> contributions to it.  That's not rigorous, and no definition can be
> rigorous.

It doesn't matter, because copyright law doesn't care about complete
programs, just about programs.

> OK.  So in your view, if on an otherwise empty CD-Rom you put the GCC
> and the new code generator together, is that a "mere aggregation"?
> Surely it must be a collective work, since these two pieces consitute a
> "collective whole", as just described above.

It is a plausible argument that this might constitute a collective
work and subject the code generator to the GPL. A careful author
would ship them separately, and best not ship GCC at all. If the
DVD were to contain a large collection of programs and include GCC
and the code generator, it would be less plausible to consider them
as an integrated whole. At this point, we would get into arguments
over fine distinctions.