dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Fri, Oct 27, 2006 at 12:50:09PM +0200, Johan Jansson wrote:
> On Fri, Oct 27, 2006 at 09:55:34AM +0200, Anders Logg wrote:
> > Yes, the script dolfin-swig is missing. I suspect Johan will add it
> > later today.
> > 
> > I have fixed the remaining bug in the assembly, so the benchmark for
> > Lagrange elements order 1-5 is running again which is a good
> > indication that the new mesh works.
> > 
> > I'll await the fix by Johan and then make a new release.
> > 
> > /Anders
> > 
> 
> Sorry, I failed adding it twice. I'm ready for a release.
> 
> PyDOLFIN and the "dolfin-swig" tool can now compile more DOLFIN
> classes than just forms and finite elements, typically Function
> (coefficients) and BoundaryCondition, which need to be efficient. It
> should be trivial to allow compilation of any interface that's
> included in PyDOLFIN, but I haven't found a clean solution. This is
> also related to having to manually input the header files in the
> PyDOLFIN SWIG interface file.
> 
> I'm not quite sure where this is heading, but it's very useful to be
> able to transparently mix Python and C++. Perhaps in the end we will
> find an existing Python module which does this better (JIT-type
> functionality is still missing). The "plast" PyDOLFIN demo shows how
> this can be used.
> 
>   Johan
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev

We could consider using "Instant": http://freshmeat.net/projects/instant/
It's a Python module that does exactly this: inlining C and C++ code
in Python.

We have been discussing moving this to fenics.org and making it a
FEniCS project (together with a couple of other projects: Swiginac,
Famms and SyFi). More on this in Delft.

/Anders