🚀 go-pugleaf

mib@geech.gnu.ai.mit.edu (Michael I Bushnell) writes:
>In article <JBW.92Jan27204933@bigbird.bu.edu> jbw@bigbird.bu.edu (Joe Wells) writes:

>   In article <MIB.92Jan27201417@geech.gnu.ai.mit.edu> mib@geech.gnu.ai.mit.edu (Michael I Bushnell) writes:

>      If you don't read the source, you are guaranteed it's not derivative,
>      so you aren't worse off than with commercial software.

>   Not good enough.  If you use a GPL-covered include file, the object file
>   is derived from it.

>You are correct; I was wrong.  The issue is *not* one of look-and-feel
>copyright.  The use of GPL-covered code is not the same thing as
>writing your own code to do the same thing.

The (IMHO spurious) idea that somehow the "user-does-the-link" approach
violates the GPL is based on either look-and-feel or interface
copyright -- which are just about the same thing. Consider the following
scenario:

1) I sell users whiz.o, which contains all the object code for my
WhizBang Word Processor. The only remaining external references are to
certain functions which happen to be provided by code in the Gnu Emacs
distribution.

2) I ship the required Gnu Emacs files (source and object) along with
whiz.o, plus a handy makefile which links the two together.

3) The user types make and, presto!, a copy of WBWP exists.

Has the GPL been violated? If so, by whom? I argue that 1) I have not
violated the GPL. No GPL-covered software has been distributed without
source. My software has never been linked with the GPL prior to my
transfer of control to the end-user; thus, it cannot have fallen under
the GPL. 2) The end-user has not violated the GPL. They have not
distributed the resulting WBWP executable (indeed, they cannot
distribute it, since whiz.o is copyrighted and not freely distributable.
Even if it were, they would run afoul of the GPL by distributing WBWP).

But, you say, this is clearly subterfuge to get around the GPL. Not so!
First, the law has nothing to say in this case about "clearly
subterfuge"; the law is the law, and there is nothing in it to prevent
the above series of events. But, for advocates of free software, there's
another reason to believe that the GPL should not cover whiz.o or the
end-user's WBWP binary.

Consider the same series of events as above, except that whiz.o needs
only some functions from libc, not from Gnu Emacs. The end-user links
not with Gnu Emacs code but code from the hypothetical GNU libc (which
might be GPL or GLPL; it really doesn't matter). Now has the GPL been
violated?

Of course not. The user could have linked with any of a number of
potential libraries which have libc functions in them. But, in the case
where whiz.o depends on GNU EMACS code, the user could have written code
to handle those calls, too. Or some third party could have provided a
compatible non-GPL library. That such a library does not exist is no bar
to its hypothetical existance, anyone is free to write a program which
duplicates any interface offered by a GPL product, unless the FSF is
asserting "interface copyright" on their code.

In claiming that a program "designed to link with GPL-covered software"
falls under the GPL, the FSF is asserting a back-door interface
copyright on the GPL code. Interface copyrights are about the same level
of vileness as look-and-feel copyrights and patents -- all three prevent
you from making compatible systems even without actually copying the
original protected work. I don't think the FSF really wants to justify
or accept interface copyrights, so I don't believe that they actually
want to have the "user-does-the-link" case fall under the GPL. It is,
however, in their best interests to scare people into believing that's
what happens.

--John

Disclaimers:
1) I work for a software monopolist now.
2) I'll be working for "even-more-vile" (TM) software monopolists soon.
3) I like the GPL, and the FSF, for some purposes; all GPL-derived code
I write is distributed under GPL guidelines (i.e. my interest in the
above workaround is hypothetical; I don't actually use it or intend to
use it in the future).

--------------------------------------------------------------------------
John L. Coolidge     Internet:coolidge@cs.uiuc.edu   UUCP:uiucdcs!coolidge
Of course I don't speak for the U of I (or anyone else except myself)
Copyright 1992 John L. Coolidge. Copying allowed if (and only if) attributed.
You may redistribute this article if and only if your recipients may as well.