[Fsatom-comm] FSAtom

[Konrad Hinsen]

On Tuesday 04 February 2003 23:53, Miguel Marques wrote:

> fairly simple to use (for what I see, I never tried it), but is heavily
> based on structures and trees. Even though Fortran 90 has structures and
> pointers, I do not think we can easily interface it with F90 (padding in
> structures, etc, is probably not consistent with C). I can do a couple of
> tests to try it out.

The mapping of F90 structures to C structures is compiler dependent. If you 
care at all about portability, forget this idea.

> I think that is a bad idea to do the I/O in Python. One can not assume
> that people are willing to use Python (after all this is a free world ;))

I don't see any need for everybody to agree on such implementation issues. 
However, I suppose natural selection will favour those who do I/O in Python, 
because is it so much easier.

> and we should decide upon a fairly universal interface (i.e. C, with
> bindings to Fortran)

You will have to choose between
1) C
2) F90
3) a common interface that doesn't use structures

I don't think any of these wil make everybody happy. The other solution is to 
write an XML parser in Fortran 90. That is certainly possible, just a lot of 
work.

Konrad.
-- 
-------------------------------------------------------------------------------
Konrad Hinsen                            | E-Mail: hinsen@cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.56.24
Rue Charles Sadron                       | Fax:  +33-2.38.63.15.17
45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
France                                   | Nederlands/Francais
-------------------------------------------------------------------------------