[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Eclipse 3.0 Running ILLEGALY on Kaffe

On Wed, 12 Jan 2005 19:08:19 -0500, Raul Miller <moth@debian.org> wrote:
> On Wed, Jan 12, 2005 at 02:58:38PM -0800, Michael K. Edwards wrote:
> > Right.  But whether it will run isn't a copyright criterion, any more
> > than whether a work of criticism will make any sense if not read
> > side-by-side with the work it critiques.
> Sure, and evidence isn't proof.
> If it can be shown that no creative effort was needed to produce some
> work, then it can probably be shown that it's not a copyrighted work.

Right.  If no creative effort is involved in combining X and Y, then
X+Y isn't copyrightable.  A copyrightable collection is one in which
there's a creative choice involved in selecting D, G, and O from among
A-Z and putting them in the order DOG.  The distinction exists so
that, for instance, republication of a collection of short stories
requires copyright license from both the editor of the collection and
the copyright holders on the individual stories -- but not if the
collection criteria are, say, "everything posted to the public
debian-short-stories list in June 2004, in date order".

If X is a code library and Y is an application that uses it, then in
an engineering sense Y depends on X.  It also contains bits of X, such
as the function names and signatures that form its API, but those bits
aren't copyrightable (per US case law such as the precedents
referenced in Lexmark v. Static Control 2004) because they're
functional rather than expressive.  Y may not work without the
behavior of and ideas in X, but that's not copyright matter either. 
So Y isn't a derivative work of X.

If combining X and Y is a mechanical, uncreative procedure -- linking,
bytecode interpretation, whatever -- then X+Y not only isn't a
derivative work, it's not a copyrightable collection.  I don't mean to
imply that writing a linker or a JIT compiler isn't a creative act,
but applying it to particular inputs isn't.

> > The GPL isn't written that way.  It invites licensees to interpret
> > "derivative work" as a term of law.  I don't think that either writing
> > a library-dependent program or linking it with that library, at build
> > time or at link time, creates a derivative work of the library, for
> > the reasons I have cited.
> I think you're agreeing with what I said in my last post.

Except that I'm also saying that writing a library-dependent program
doesn't create a derivative work under copyright law, and that there's
no contributory infringement going on, either.

US and UK courts (and doubtless others) have placed rather strict
limits on the abuse of statutory property rights such as copyright and
trademark to restrain legitimate competitive activity.  That's where
affirmative defenses such as "fair use" and "reverse engineering" came
from, along with criteria such as "functional vs. expressive" and
"ideas vs. expression" that limit the reach of copyright.  (All of
these were judge-created before they were enshrined in statute.)

There's no a priori reason whey they wouldn't also apply those limits
to Free Software licenses, and the precedents I've cited seem to
suggest that stretching the GPL across linking boundaries goes over
the limit.

- Michael

Reply to: