RE: Distribution of non-GPL models with GPL Software (Was: [Qucs-devel] Closed bugs)
David Bremner wrote:
[snip]
> Speaking only for my self, I don't see a problem with shipping DFSG-free [0]
> licensed data with GPL code. As long as the code runs fine without the models in
> question, I guess it is pretty clearly not linked.
>
> [quote of proposed license]
>
> >> Any changes beyond the above (which may affect performance or
> >> function of the model) are permitted, but must be distributed as a
> >> "patch file" alongside the original, unmodified spice model.
>
> At first I thought this completely in the spirit of DFSG 4, and thus fine from a
> Debian policy point of view. But on second thought I'm not so sure. I'm not sure
> exactly how to apply "The license must explicitly permit distribution of software
> built from modified source code." to this case.
That's the confusing part -- what is defined as "software" and "source code".
In this case - Qucs is the software. Any contributions we did for the Qucs source code,
would be released under the Qucs license (GPL).
Models, which care not compiled into Qucs, are not "source code" of Qucs, and are
more like "data files" along side Qucs, which Qucs operates on. Qucs runs fine without
them, and the models are only included as a convenience to the user.
> I guess it would be OK if the
> license permitted modified files as long as the original and a patch were also
> included, although that makes the patch a bit redundant.
It's hard for us to do that - since our models go through a pretty rigorous QA process.
People modifying the model to make one aspect better, and another worse may be OK
for that specific user, but not everyone. This is why we thought a original + patch would
work better.
> Maybe somebody else has more experience with e.g. data sets and can suggest
> an existing license which might be suitable. This is almost always preferable to
> creating a new license.
Yeah, I looked around for some examples - the only thing I found was gnuplot.
http://gnuplot.cvs.sourceforge.net/gnuplot/gnuplot/Copyright?view=markup
The reason it's not 100% appropriate, is it describes "software". The model is not software (IHMO).
The model is used to direct a computer simulation (in this case Qucs, "the software").
The model can't do anything by itself, or be compiled into anything useful.
> [0] http://www.debian.org/social_contract#guidelines
Reply to: