Subject: Re: balance
From: "Karsten M. Self" <>
Date: Mon, 27 Sep 1999 16:50:33 -0700

Recipients:  non-FSB recipients dropped.


L. Peter Deutsch wrote:
> > Nahhhh.  Copyright law includes the concept of "fair use".  A single word
> > portion isn't going to bring down the GPL on you.  Remember, the GPL is a
> > license to use a copyrighted work, so if it's not covered by copyright
> > law, the GPL doesn't apply.

I'd tend to agree with Russ's comment here -- the GPL doesn't apply
where copyright doesn't restrict.
> Does "fair use" extend to a large table all of whose elements are constants
> #defined in a copyrighted header file?

Might, might not.  First, the law isn't stated in black and white. 
However, mitigating conditions might include the originality of the
original table, the extent to which it is expressive (covered by
copyright) vs. functional (not), whether the table properties might be
discoverable by reverse engineering but not be capable of being stated
in a different form, the extent to which the values themselves are or
are not original creations of the copyright holder of the file they are
contained in (say, POSIX definitions).
> Does "fair use" extend to allowing me to distribute patches for GPL'ed code
> without putting the patch file under the GPL?

I believe there's a concept of contributory infringement which comes
into play here, and that this was one of the issues with the Qt
license.  Even though the patch file itself isn't a derived work, the
patched sources are, likewise the compiled and linked object code
(either on disk or in memory).

> > Nahhhh.  You can't copyright lists of things.
> Sometimes you can't copyright lists of things that existed before and
> independently of the list.  That isn't the case here.

Feist v. Rural Telephone (or Rural Electric?) -- simple compilations of
fact are not copyrightable.  The DMCA has specific provisions applying
to copyright of databases, however.

> > A header file is just a list of entry points, and constants.
> A header file can be copyrighted just as any other piece of code can.  A
> header file is the expression of a software design and is the result of
> creative work: it's not just an aggregation.  If a list of command names can
> be copyrighted (which it can), then a collection of command names, argument
> names, argument types, structure definitions, and parameter values embodying
> a software design can surely be copyrighted.

It's important to realize that a copyright symbol, statement, or even
registration, isn't sufficient to establish copyright -- the Copyright
Office (in the US) doesn't pass judgement of copyrightability of
registered works.  Copyright itself is ultimately decided in the
courts.  For expressive, original works, it's generally pretty cut and
dried.  Special cases (boat hulls, databases, architectural features)
and exemptions (fair use and others) make for the interesting minority.

As far as simple lists, Lotus v. Borland held that command menus were
not a copyrightable feature of a program (Borland had copied Lotus 123's
menu command structure).  I'd have a hard time with a STRUCT as well --
how is it like a recipe and how is it like a story (recipies are not
copyrightable -- recipie books with photos and narrative are).  Selden
v. Baker -- accounting forms are functional, not expressive, and are not
copyrightable.  Sega v. somebody (Accolade?) -- a small number of bytes
(14?) which had to be copied literally and was required to run a game
cartridge in a console, was not a copyrightable portion of a work. 
While a given definition of a file or data structure might be
copyrightable, the structure itself is an idea, and cannot be subject to
copyright.  The extent to which the structural definition in software is
an expression, and the extent to which it is a non-expressive
description of the structure, I'm not sure.

Karsten M. Self (
    What part of "Gestalt" don't you understand?

SAS for Linux:
Mailing List:  body "subscribe sas-linux"