fenics team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Tue, Apr 05, 2011 at 11:49:05AM +0200, Harish Narayanan wrote:
> On 4/5/11 8:39 AM, Anders Logg wrote:
> > We're making good progress with collecting the copyright forms and
> > should soon be able to make the switch to LPGL.
> >
> > A couple of points I'd like to make:
> >
> > 1. When someone submits patches, maintainers first need to ask
> > contributors to sign the two forms. Otherwise, we risk having to run
> > after people we don't know to sign the forms later.
> >
> > 2. FEniCS Apps should have the exact same license as the rest of the
> > code, simply because that enables copying of code from Apps to Core.
> > It is natural (and desirable) that some of the code developed as part
> > of an App moves into DOLFIN if it's found that code may be useful to
> > other projects.
> >
> > This means CBC.Solve needs to either use the LPGL, or, if Harish still
> > objects, be removed from FEniCS Apps or CBC.Twist removed from
> > CBC.Solve.
>
> I understand and mostly agree with what you are saying.
>
> The only strong opinion I have is about cbc.twist. Not to be difficult,
> but I genuinely feel its goals---being a test-bed to learn and educate
> others about mechanics---are best served if any further projects built
> upon it are developed in an open fashion. To enforce this, I would like
> it to remain GPL.

I understand, but it's a complication if we can't copy code between
the projects.

Other opinions?

> I think the easiest solution here is that I split cbc.twist from
> cbc.solve. This allows the rest of cbc.solve to then be an LGPL FEniCS
> App, and I have no other qualms about signing the license change forms.

So it would be an add-on to CBC.Solve? It would be a bit unfortunate
if we'd have to remove it from CBC.Solve. But perhaps it would be
fairly easy to implement a new hyperelasticity solver.

--
Anders