Re: Eclipse 3.0 Running ILLEGALY on Kaffe

[Walter Landry <wlandry@ucsd.edu>]

Brett Parker <iDunno@sommitrealweird.co.uk> wrote:
> On Wed, Jan 19, 2005 at 12:52:29PM -0500, Brian Thomas Sniffen wrote:
> > Michael Poole <mdpoole@troilus.org> writes:
> > 
> > > As has been settled on this list, Eclipse is not a derivative of Kaffe
> > > and does not contain any copyright-protected portion of Kaffe.  It is
> > > possible to claim that "Eclipse+Kaffe" is a work based on Kaffe, but
> > > by the same argument, "Debian" is a work based on Kaffe, and the
> > > rational interpretation is that both cases are mere aggregation.
> > 
> > It seems to me that "mere aggregation" must be the smallest idea that
> > is still aggregation.  For example, Emacs and Vim are merely
> > aggregated in Debian.  wget and openssl are not merely aggregated,
> > because there's more going on there.  It's not necessary to look in
> > great detail at what *is* going on there -- it's enough to say that
> > there is more there, so it's not merely aggregation.  It's aggregation
> > and something else.
> 
> wget and openssl are linked, openssl is a build depend of wget, it is
> very much required to compile it. So, yes, it is not mere aggregation.

What if there was a package wget++ that communicated with openssl
entirely through system() or exec() calls?  It would construct
appropriate input and parse openssl's output.  Would that constitute
linking?  It ends up using all of the same code as the directly linked
version.

If it is not linking, why couldn't you do this with all GPL'd
libraries?  You could write a GPL'd wrapper around a library, and just
use the wrapper with exec().

In essence, why does using exec() suddenly break the chain, while a
linker or classloader does not?

This is not a outlandish example.  Front-ends to CVS do this, more or
less, because CVS does not have a library interface.

Regards,
Walter Landry
wlandry@ucsd.edu