Subject: Re: Stallman vs Wind River on the GPL
From: "Jonathan S. Shapiro" <shap@cs.jhu.edu>
Date: Fri, 02 Mar 2001 08:41:41 -0500

kragen@pobox.com wrote:
> 
> "Jonathan S. Shapiro" <shap@cs.jhu.edu> writes:
> > My point is only that (a) this issue MUST be clarified in the next
> > version of GPL, and (b) it must be clarified in a way that accepts
> > current common practices (e.g. Torvalds' position vs. RMS's position) or
> > we may face a schism.
> 
> Code the copyright
> holder licenses under the GPL is licensed under the GPL; other code
> isn't.  Period.

The question at hand is not the choice of license, but the constraints
on combination that result from this choice. The dynamic linking issue
is a pivotal issue here, and the application of GPL to dynamic linking
has been interpreted both ways by different communities. Since the
community is clearly divided on this point, GPL^n must somehow straddle
the boundary or risk a schism. I will be interested to see how Richard
and the FSF group working on GPL handle this (in fact, Richard, I'ld be
interested to review a GPL3 draft if I might).

This is all going to get even more confusing next year when EROS is
really runnable -- EROS is a system where RPC has very much the same
flavor as dynamic linking, and we have already had to fumble with
defining what "combined work" means in this context. Our fumblings,
which are better than nothing but not great, can be found at
http://www.eros-os.org/legal/terms.html

Jonathan