On Monday 07 October 2002 13:37, Thomas Bushnell, BSG wrote:
> Auke Jilderda <auke@jilderda.net> writes:
>
> > - First, the boundaries of the GPL are unclear. Exactly what does the
> > term "derived work" mean, does the license propagate across static
> > linking, dynamic linking, IPC, or even socket communication? This
> > unclarity is a risk for companies and, consequently, they take a
> > cautious approach, staying on the safe side by not linking their
> > proprietary software (that contains their business value) to GPL
> > software. In other words, the unclarity in the GPL license causes
> > that software to be used a bit less than had it been clear about its
> > boundaries.
>
> "Derived work" is a well-established term in copyright law. The
> reason the GPL doesn't give a local definition is because it reaches
> exactly as far as the normal meaning of a derived work. The
> boundaries of what is a derived work are *exactly* the same,
> therefore, as for any other copyright program.
It's well-established, but that doesn't mean it's well-defined. The
edges are defined only by litigation, and sometimes redefined by litigation.
Hence the boundaries of the GPL are unclear (and you subsequently
noted a case (patches) where that ambiguity exists).
Lynn