fenics team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Thu, Nov 16, 2006 at 12:40:50PM +0100, Garth N. Wells wrote:
> 
> 
> Anders Logg wrote:
> > On Thu, Nov 16, 2006 at 09:56:12AM +0100, Garth N. Wells wrote:
> >> Anders Logg wrote:
> >>> I'm suggesting two new projects to be added to the web page as FEniCS
> >>> projects:
> >>>
> >>>  1. SyFi by Kent-Andre Mardal
> >>>
> >>>  2. UFC by Alnæs, Langtangen, Logg, Mardal, Skavhaug
> >>>
> >>> SyFi is described in detail here:
> >>>
> >>>     http://folk.uio.no/kent-and/software/SyFi/doc/SyFi.html
> >>>
> >>> SyFi can be seen as a combination of FIAT and FFC using symbolic
> >>> manipulation through GiNaC in C++ and Python.
> >>>
> >>> UFC (Unified Form-assembly Code) is a specification of a common
> >>> format for the code generated by FFC, SyFi and other form compilers.
> >>>
> >>> Are there any objections to adding these two new projects?
> >>>
> >> Sounds good to me. UFC will provide a nice link between different projects.
> >>
> >> Will both projects be hosted on www.fenics.org? For the coherence of
> >> FEniCS, I think that this is preferable.
> > 
> > Yes, that's the plan.
> > 
> > Kent has made effort to FEniCSize SyFi: using fenicsmanual.cls for the
> > manual, conversion from Subversion to hg. There will be a web page,
> > mailing list syfi-dev@xxxxxxxxxx etc.
> >
> 
> Great. It's important that all the sub-projects have some common
> "FEniCS" elements to present a unified face, otherwise it will become a
> loose collection of unrelated projects (a mini sourceforge).

I agree.

We don't have any written policies for which projects might be
considered for inclusion, but this is something we should discuss.

/Anders

> Garth
> 
> 
> > /Anders
> > 
> > 
> >> Garth
> >>
> >>> We also plan to add another new project UFL (Unified Form Language)
> >>> that specifies (and implements) a common input language for SyFi and
> >>> FFC. At that point, SyFi and FFC will function similarly - the same
> >>> input gives the same output (but with different optimizations) - and
> >>> we may then consider a merge in one form or the other. The methodology
> >>> for the code generation can then be controlled by a simple flag to the
> >>> compiler: tensor representation, symbolics or quadrature.
> >>>
> >>> /Anders
> >>>
> >>> PS: As we discussed in Delft, there are more projects in the pipeline
> >>> but we can discuss these projects separately when they are ready for
> >>> inclusion.
> >>> _______________________________________________
> >>> 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
> > 
>