F2PY problems

[Siegfried Gonzi]

Fernando Pérez wrote:

> Are you on the scipy-users mailing list? Pearu (f2py's creator) hangs out
> there and he's extremely helpful and responsive, I'm sure you'll have better
> luck there than here, as f2py is a rather specialized tool these days.

The proble actually is that Pearu does not have any access to a windows
machine. He cannot help me; he gave me the tip to come up with my
problem on the Python mailinglist (though I post it here because the
mailinglist does not have a digest option and I do not want to stress my
mailbox).


I detected the usefulness of Fortran again; I always thought that there
exists no free Fortran 95 compiler; but there is one available from NAG
called F. It is evident that most of the scientist would throw away C or
C++ in order to use Fortran 90/95 (the problem is often the scarceness
of free Fortran 95 compilers; in the past I was inclined to give C/C++ a
try especially with respect to a developing country; but there is
nothing out there which can more hurt a mans brain than C/C++).

That said. 

I got the advice to try to force the system to use mingw32-support. The
scipy-distutils is included with the F2PY distribution, but when I force
the use of mingw32-support I get the error message that "lib2def" cannot
be imported. The ming32-support.py file tells me that "lib2def" is
somehow connected with weave. But the F2PY distriubution does not
mention any weave libraries. I then deleted the weave part but it is
clear I get the error message that gcc will possibly fail.

I have to underline that Mingw is installed on my Windows XP machine and
I use it successfully (otherwise I couldn't use the Fortran 90
compiler). I have further to emphasize that I can use SWIG without
problems; at least I made an external wrapper function a few weeks ago;
okay SWIG relies on Microsoft Visual C++.


Is there an alternative to F2PY?

S. Gonzi