David A. Case wrote:
> On Tue, Mar 03, 2009, Joe Krahn wrote:
>
>> I was discouraged about Fortran standards until F2003 came out. Among
>> other things, it defines a standard method of directly interfacing to C,
>> for both calling C and being called from C. This makes it easy to
>> utilize common C libraries,
>
> I tend to agree with Bob here: we've never had troubles interfacing to C
> (unless, I guess, you want to pass Fortran strings or records, or things
> like that). As example, we have both Fortran and C (and assembler)
> versions of the blas and lapack libraries, and they can be (and are) called
> from both C and Fortran driver routines.
>
> Is there a specific thing in Fortran2003 that you want to use?
>
> ...dave
>
Calling C works now, but means having to be aware of differences among
different compilers. F2003 just makes it easier, and means I don't have
to figure out AMBER's set of CPP macros, and it means that AMBER needs
less compiler-specific interface work.
I agree that it is not worth trying to use in shared production code if
most users don't already have F2003 compilers. I have started using it
because the Fortran compilers I use support it now. I just didn't know
how common F2003 support is, but it sounds like it is still way too early.
There are some other nice features besides C, like being able to get a
text error message from a compiler-specific IOSTAT value, and IEEE
functions for dealing with NaN and Inf values. A few of the more useful
F2003 features are already widely available, like allocatable members in
derived types.
Joe
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Mar 06 2009 - 01:24:30 PST
