Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!usc!brutus.cs.uiuc.edu!coolidge
From: coolidge@brutus.cs.uiuc.edu (John Coolidge)
Newsgroups: comp.sources.d
Subject: Re: What does free mean.
Message-ID: <1990Mar12.182829.6108@brutus.cs.uiuc.edu>
Date: 12 Mar 90 18:28:29 GMT
References: <1151@mtxinu.UUCP>
Sender: news@brutus.cs.uiuc.edu
Reply-To: coolidge@cs.uiuc.edu
Distribution: usa
Organization: U of Illinois, CS Dept., Systems Research Group
Lines: 130

frk@mtxinu.COM (Frank Korzeniewski) writes:
>In article <1990Mar12.060117.2305@brutus.cs.uiuc.edu> I write:
>#frk@mtxinu.COM (Frank Korzeniewski) writes:
>#>Kyoto Common Lisp requires that you sign a license to get their free
>#>lisp system with C source.  You have access to the source before they
>#>recieve the license.  This license requires, among other things,
>#>that you get permission from them before you can distribute any
>#>modified copies to others.
>#
>#This is a non sequitor. If you have access to the source before you give
>#them the license, then there's no way they can force you to file the
>#license.

>Listen carefully.  The source is available through anonymous ftp.  You are
>on your honor not to steal it.  When you sign and mail the license you
>can access the source, without waiting for them to recieve and process it.
>If you have a problem with this procedure, don't take it up with me.

My previous comments on licenses were partially goofed up, sad to say
(it's what I get posting at that time of day :-) ). However, it would
appear, from this discussion of the KCL setup, that what they're doing
is offering a non-binding license agreement. It's entirely up to each
user to decide whether or not they want to follow the license. Such a
license is not legally enforceable. I don't "have a problem" with this,
I'm just pointing out that "on your honor" isn't a legally binding
form of protection.

>#>There are no such complications with the GNU assembler (GAS) or the GNU
>#>loader (GLD).  Nor with all the numerous utilities.
>#
>#libg++ _is_ explicitly under the copyleft, and anything linked with it
>#must (if released) be covered by the copyleft. In fact, the FSF claims
>#even stronger protection for libg++ (and other Gnu code): any package
>#designed to link with libg++ or the source to a Gnu tool _must_ be
>#covered by the copyleft. Furthermore, any libg++-compatible library
>#must also be covered by the copyleft. This was all posted to gnu.g++
>#and/or gnu.misc.discuss (by the authors of the respective packages).
>#I have major doubts about the both the legality and the ethics of
>#this stance, but it's the current position.

>Fine, I forgot about libg++ because I do not use it.  I have no problems
>with the legality or the ethics of their position.  Like you, I am not a
>lawyer either.

My problem with legality (see below) is that the FSF is illegally
tampering with my copyright to my code. My problem with ethics is that
the FSF has come out against proprietary user interfaces many times, but
yet they're trying to argue for proprietary library interfaces. I
consider this unethical behavior and bad for programmers in general.

>#What if AT&T libc carried a similar restriction (namely, that any code
>#linked with it would fall under the AT&T copyrights)? In that case,
>#building GCC would place it under the AT&T copyright. In other words,
>#your "it's not a problem, there are other sources" attitude works for
>#the small case, but breaks badly in the larger scheme of things.
>#

>This comment of yours makes absolutely no sense.  I see nothing
>breaking badly in your statements.

Hmm. I think it's rather clear. Once again, slowly. What if AT&T's libc
carried with it terms like those on libg++ (link with this library and
you agree that your source code is covered by the AT&T copyright and
cannot be redistributed except with express permission of AT&T and under
AT&T's terms --- license agreements, fees paid to AT&T, etc). Had this
been done, there never could have been an FSF. Of course, Unix as we
know it wouldn't exist either, but that's beside the point.

What I said "breaks badly in the larger scheme of things", what I meant
was fairly simple. You show that the FSF's policy of "link with my stuff
and your stuff is covered by my rules" is harmless in a few small cases.
I show an example of where the same policy could do great damage. In this
way I hope to show that the policy itself is fatally flawed, since it
permits great evil to be done. Simple, no?

>Noone forces you to link with the GNU code.  Noone forces you to jump off
>a cliff either.  You do both at your own risk.  You are aware of the
>consequences of both (I hope).  Others of us happen to agree with the
>FSF position.  Given our legal system, they either have the restrictions
>they do, or people can subvert their intentions to keep their code free.
>Our legal system gives us choices like this.  Take it up with the
>government if you disagree.

The problem is that the FSF's restrictions are almost certainly illegal.
Linking with another piece of code does _not_ force other code to be
covered by the same copyright. As someone pointed out in gnu.misc.discuss
a short while back (someone who _was_ a copyright lawyer), the link-
therefore-copyleft argument violates the section in the copyright act
which governs transfer of copyright. I already have the copyright to
my code _before_ I link with the FSF's code, and they already have the
copyright to their code. After the link, it is equally as silly to say
that my code falls under their copyright as it is to say that their code
falls under my copyright, or that either set of code falls under the
AT&T copyright (since I did, after all, link both sets of code with an
AT&T-derived libc). The FSF still owns their code and controls the way
it can be distributed (with full source). I still control my code and
control the way _it_ can be distributed (with full source, maybe; maybe
with _no_ source), and AT&T still controls _their_ code and its rules
of distribution (no source, of course).

If you don't follow this line of reasoning --- if you say that, once
something is linked with FSF code, full source distribution of the
entire package is required, then the binary-inclusive distributions
of gnu tools (yes, _even_ those done by the FSF) are in violation of
their own copyleft. You see, they contain bits of AT&T-derived libc and
other libraries, and those who pass them on _cannot_ give out source to
those libraries.

>#Copyright 1990 John L. Coolidge. Copying allowed if (and only if) attributed.
>#You may redistribute this article if and only if your recipients may as well.

>Your signature carries the GNU philosophy.  The intent of both is to retain
>the rights.  I find it amusing that you object to the FSF.

No. If I was following the FSF's philosophy, my signature would require
that your article _also_ be freely distributable. I do not --- I merely
require that 1) you give me attribution, and 2) that my article itself
(and not any amendments to it) be freely distributable. It's _exactly_
the same as my views on free software.

In any case, I hardly "object to the FSF". I use FSF tools all the time,
and I maintain and submit bug fixes to their code. What I object to is
silly reasoning that would lead to disaster if applied to anyone but the
FSF. What I object to is the occasional bit of FSF hypocrisy wrt
proprietary interfaces. And what I object to, most of all, is the idea
that it is legal, ethical, and legally enforcable for some strange sort
of compilation copyright to come out of the act of linking together
disjointly copyrighted pieces of code.

--John