Subject: Re: "Derivative Work" for Software Defined
From: Ian Lance Taylor <>
Date: 15 Jan 2003 12:24:09 -0800

"PETERSON,SCOTT K \(HP-USA,ex1\)" <> writes:

> "The only issue is whether G+H is a derived work of G, and that seems
> obvious."
> I think there is an additional issue, and one for which the resolution is
> not necessarily obvious: to what sort of combinations of G and H does the
> condition in section 2 of the GPL extend? In other words, what sorts of
> things are G+Hs rather than just G aggregated with H?

I don't see why that is relevant for the example which you presented.
Can you explain?

I must admit that I don't know what you are driving at in this series
of messages.  That part of section 2 of the GPL seems to me to be
aimed at clarifying the GPL authors' interpretation of what derived
work, or ``a work based on the program,'' means in the context of
computer software.

I think it's important to note that that part of section 2 doesn't
actually claim any rights over work which is not derived from the
GPL-covered work.  The only point of that text is to clarify that the
authors of the GPL don't intend certain cases of distribution to be
covered by the GPL.  And they say what those cases are: work which
``can be reasonably considered independent and separate works in

I would say that if you have a G+H case which is not, in law, derived
from GPL-covered G, then the GPL does not apply (this test is not
currently useful because the law does not exist).  I would also say
that if you have a G+H case which is, in law, derived from GPL-covered
G, but which ``can be reasonably considered independent'' of G, then
the GPL does not apply.

So now that I've said that, I see that you could be asking how to
decide those cases in which G+H is, in law, derived from G, but which
at the same time might possibly ``be reasonably considered
independent'' of G.

If that is your question, my personal answer would be two-fold.
First, we don't know if there are any such cases, because we don't
know what the law says about whether G+H is derived from G.  Second,
if we did know the law, I doubt there would be any such cases; I
suspect that if any G+H were, in law, derived from G, then nobody
would reasonably consider such a G+H to be independent of G.

So my personal feeling is that you may be trying to clarify a category
which does not exist.  I interpret that part of section 2 of the GPL
as the GPL author's attempt to provide guidance in the absence of law,
not in replacement of law.

license-discuss archive is at