Subject: derived works (Re: Who holds the copyright?)
From: "Stephen J. Turnbull" <>
Date: Thu, 18 Nov 1999 19:33:10 +0900 (JST)

>>>>> "Alessandro" == Alessandro Rubini <> writes:

    >> I've asked before (but still not gotten an answer I understand)
    >> where the `exec' function got the power to define a `work'.

    Alessandro> Everybody agrees that system call interface is a
    Alessandro> work-boundary. Interpretations differ about dynamic
    Alessandro> loading. Using a public interface via dynamic linking
    Alessandro> is considered acceptable by Linus and not acceptable
    Alessandro> by Richard.

Well, AFAIK what Linus says or Richard says is irrelevant.  It's not a
matter of what people on FSB or in the free software community
"accept" or "agree," at least not according to what Jonathan Shapiro
wrote (<> on Oct 17 on FSB):

    JS> First, the meaning of "derived work" is a technical term
    JS> defined in the legal community with a fair degree of
    JS> precision.  It means a work of authorship incorporating more
    JS> than a certain specifically-defined minimum amount of content
    JS> from some other work of authorship.

Granted, RMS _is_ a reliable authority on such legal issues (subject
to the caveat RMS-INAL) in the sense that he does his homework and
(what a concept!) actually asks experts about such things.  On the
other hand, while I believe RMS would never lie if told by his lawyer
"dynamic linking is OK under applicable law and there is nothing you
can do about it," I also believe that if his lawyer told him
"following cases such and so there's a reasonable argument that
dynamic linking creates a derived work" he would _never_ publically
express doubt that said reasoning would be upheld if tested in court.
Certainly not in the same breath as telling Peter Deutsch that he
cannot include GNU readline with Aladdin Ghostscript.

Honesty is a virtue which RMS possesses, but I can't see him going out
of his way to encourage people doing things he tells them to stop
doing to test his authority in court!  So I don't trust RMS's
published statements to be a good estimator of what would happen if
(say) the FSF should actually decide to sue Peter Deutsch (or Red Hat
or Caldera!) for including a (dynamically linked) interface to GNU
readline in Aladdin Ghostscript.

Now, I infer (perhaps incorrectly) from Jonathan's statement that
"incorporating content from another work" is also a technical term
defined in law, _not by private contract, ie the GPL or the common
understanding of the intent of the GPL_.

And that is my question:  does applicable law define whether a program
which (a) exec's (b) is dynamically linked to (c) is statically linked
to a second program is a derivative work of the second program?

If no, then according to law what does define this?  If yes, which is
it (a, b, c), and what's the reasoning?  (I'm willing to believe that
in the end it really does come down to "what RMS says," but I'll take
more convincing than "that's what RMS has been saying for 20 years.")

University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
What are those two straight lines for?  "Free software rules."