On Fri, Jan 04, 2002 at 06:03:30PM +1100, Hamish Moffatt wrote: > On Thu, Jan 03, 2002 at 10:43:48PM -0800, Thomas Bushnell, BSG wrote: > > tb@becket.net (Thomas Bushnell, BSG) writes: > > > Yes, it is different. One is a program making callouts to a different > > > entity, the kernel. The case we were talking about is that of library > > > linking. > > I should add here that it is relevant that the callouts to the kernel > > are callouts to an interface which is defined as "not making things a > > combined derived work", which is not normally the case for a library. > > It is relevant and important here that the authors of the kernel > > intend that understanding of those callouts. > What is the definition of a "callout"? > Why is it so different to a published library function? > Apart from convenience of argument, that is. I think you're overlooking the fact that in the case of a GPL library, the publishing of the interface is ALSO done under the GPL. Are you using GPL header files when compiling? Then the binary output is a derived work of the GPLed library. This is not to imply that someone who reverse-engineers a GPL header file can necessarily link against a GPL library without also being bound by its license terms; it only shows that in the usual case, there's clear support for the claim that library linking creates a derived work. Steve Langasek postmodern programmer
Attachment:
pgp5E5LQn1muY.pgp
Description: PGP signature