← Back to team overview

fenics team mailing list archive

Re: New project?

 

I don't think it is necessary for all projects to be "pure FEM" codes.
I imagine it might be useful to separate out a few libraries from
DOLFIN, like the parameter system and the log system. It would then be
practical to place these libraries as separate projects on fenics.org.

Other opinions?

/Anders


On Tue, Jan 09, 2007 at 10:18:31AM +0100, Martin Sandve Alnæs wrote:
> I think Garth is right, Instant can get more exposure at SF, from
> people who would never be interested in fenics.
> 
> On 1/8/07, Kent-Andre Mardal <kent-and@xxxxxxxxx> wrote:
> > Hi,
> >
> > It is up to you. Instant is currently
> > a sourceforge project and I am happy to let it stay there.
> >
> > However, the only "real" use of Instant that I am aware of
> > is in finite element computations (it has been used with Diffpack and is
> > in use with PyCC/SyFi). It is in combination with code generation in
> > Python that it is useful. The reason Instant has been seperated
> > out of the projects PyCC and SyFi is that I think that separation is a
> > nice design goal if it is "natual".
> >
> > So, whether it "fit in" or not, I don't know. It is in use with SyFi,
> > but I don't think that is a strong reason for putting it out
> > on fenics. SyFi uses other packages which are not fenics packages.
> >
> > Kent
> >
> > man, 08,.01.2007 kl. 14.51 +0100, skrev Anders Logg:
> > > We have never decided on what the requirements should be on a FEniCS
> > > project. I agree that we should not become a SourceForge for
> > > scientific computing (it's more like GNOME or the Mozilla project).
> > >
> > > What should the requirements be? A very loose requirement would be
> > > that it should "fit in", something that
> > >
> > >   1. provides new functionality by using some of the current components;
> > >
> > > or
> > >
> > >   2. implements functionality used by some of the current components;
> > >
> > > Instant fits in under (2). We currently have dolfin-swig in DOLFIN
> > > that does something similar to what instant does and when adding a
> > > dependency from DOLFIN to Instant, it's practical to put Instant on
> > > fenics.org.
> > >
> > > But let's discuss what the requirements should be.
> > >
> > > /Anders
> > >
> > >
> > > On Mon, Jan 08, 2007 at 02:18:33PM +0100, Garth N. Wells wrote:
> > > > Is FEniCS the most appropriate place for Instant? It's a generic tool
> > > > (not PDE/numercal analysis/modelling specific) so being part of FEniCS
> > > > might not help Instant in terms of exposure. Sourceforge is probably
> > > > better(?). Also the theme behind FEniCS components becomes diluted. clear.
> > > >
> > > > Garth
> > > >
> > > > Anders Logg wrote:
> > > > > I suggest that we add Instant as a new FEniCS project:
> > > > >
> > > > >     http://pyinstant.sf.net/
> > > > >
> > > > > This project was discussed briefly in Delft. For those who missed the
> > > > > introduction, Instant is a simple tool for inlining C/C++ in Python.
> > > > > Here's an example:
> > > > >
> > > > >   >> from instant import inline
> > > > >   >> f = inline("double f(double x, double y) { return x+y; }")
> > > > >   >> f(2, 3)
> > > > >   >> 5
> > > > >
> > > > > We currently have something similar in PyDOLFIN but could use Instant
> > > > > instead.
> > > > >
> > > > > Any objections?
> > > > >
> > > > > /Anders
> > > > > _______________________________________________
> > > > > FEniCS-dev mailing list
> > > > > FEniCS-dev@xxxxxxxxxx
> > > > > http://www.fenics.org/mailman/listinfo/fenics-dev
> > > > >
> > > >
> > > >
> > > > _______________________________________________
> > > > FEniCS-dev mailing list
> > > > FEniCS-dev@xxxxxxxxxx
> > > > http://www.fenics.org/mailman/listinfo/fenics-dev
> >
> > _______________________________________________
> > FEniCS-dev mailing list
> > FEniCS-dev@xxxxxxxxxx
> > http://www.fenics.org/mailman/listinfo/fenics-dev
> >
> _______________________________________________
> FEniCS-dev mailing list
> FEniCS-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/fenics-dev


Follow ups

References