Re: can one build commercial applications with latest gnat and other licenses related questions...

[M E Leypold <development-2006-8ecbb5cc8aREMOVETHIS@ANDTHATm-e-leypold.de>]

Michael Bode <m.g.bode@web.de> writes:

> Jeffrey Creem <jeff@thecreems.com> writes:
> 
> > If you take something like the GNU scientific library and dynamically
> > link to it would be seen as trying to work around the GPL and not seen
> > as trying to meet the intent of the license since if that is what the
> > authors wanted, they would have gone with the LGPL for that library.
> 
> Then I would like to hear what the FSF thinks 'using a library' is as
> opposed to 'derive work from a library'.

Good question. The FSF (whose point of view actually has zero impact
on the discussion about GPL software from ACT, since it's not the FSF
forcing ACT in any way to use GPL ...) has, with regard to the GPL,
probably the opinion that linking means "derive a work". 

I, personally think, that has historic reasons. When the GPL was
conceived, dynamic linking hardly existed. The LGPL has been
formulated as a response to situations like those that arouse with
bison: Essentially we have program generator, but since it introduces
pieces of own code (read runtime) into the generated code, that leads
to contamination of everything produced with bison.

The FSF has IMHO not addressed the issue of merely using an interface,
which is already present at the target system (like the libc) not even
in GPL 3. I think, that should constitute a case where the executable
doesn't fall under GPL automatically. Saying that using a machine
readable interface description (like C header files) already makes a
derived work is obviously absurd.  Some rudiments of that kind of
thinking can actually be found in the exception to distribution clause
(don't have do distribute stuff already present at the target system).

Some other examples if anybody actually wants to argue against:

  - I taught a number of people Software Engineering. My texts were
    available under the FDL (Free documentation license). Should I
    have made GPL from that? Everything those people produce in future
    is a derived work of my knowledge.

  - When compiling on a linux system, The libc headers indirectly
    include (at compile time) headers from //usr/include/linux which
    AFAIS come from the Linux source tree, which is under GPL. Does
    that mean, every excutable compiled on a Linux system is under
    GPL?

I could continue her, but (wether there is provision for that in the
GPL or not), I'd suggest that for sanities sake there should be
limits: "Use" vs. "Creating a derived Work". For everybody who has
studied computer since, the difference should be clear: Interface
vs. Implementation. That is not the GPL philosophy, but that is why
the author of any library or tool has the option to put runtime or
library under LGPL or add a linking exception. 

Those who don't want that, are free (!) to not do that. They are even
free not to release anything BTW. But if they choose GPL over LGPL or
GMGPL or whatever I don't want to hear all that stuff about "more
freedom". It's not, not even in the sense of the GNU manifesto
(reducing dependencies ...!). It's an attempt (completely legal) to
lock a library away from commercial use.

I, personally, doubt the value of forced contributions of free
software to the community. Have a look to the BSD communities for a
comparison.

And still I'm for free software and for the GPL (in the right
context). Mind that.


Regards -- Markus