Here is a simple example that will let you read Fortran namelists from C. I used the namelist file that you provided in the question, input.txt
.
Fortran subroutine nmlread_f.f90
(notice the use of ISO_C_BINDING
):
subroutine namelistRead(n,m,l) bind(c,name='namelistRead')
use,intrinsic :: iso_c_binding,only:c_float,c_int
implicit none
real(kind=c_float), intent(inout) :: n
real(kind=c_float), intent(inout) :: m
integer(kind=c_int),intent(inout) :: l
namelist /inputDataList/ n,m,l
open(unit=100,file='input.txt',status='old')
read(unit=100,nml=inputDataList)
close(unit=100)
write(*,*)'Fortran procedure has n,m,l:',n,m,l
endsubroutine namelistRead
C program, nmlread_c.c
:
#include <stdio.h>
void namelistRead(float *n, float *m, int *l);
int main()
{
float n;
float m;
int l;
n = 0;
m = 0;
l = 0;
printf("%5.1f %5.1f %3d\n",n,m,l);
namelistRead(&n,&m,&l);
printf("%5.1f %5.1f %3d\n",n,m,l);
}
Also notice that n
,m
and l
need to be declared as pointers in order to pass them by reference to the Fortran routine.
On my system I compile it with Intel suite of compilers (my gcc and gfortran are years old, don't ask):
ifort -c nmlread_f.f90
icc -c nmlread_c.c
icc nmlread_c.o nmlread_f.o /usr/local/intel/composerxe-2011.2.137/compiler/lib/intel64/libifcore.a
Executing a.out
produces the expected output:
0.0 0.0 0
Fortran procedure has n,m,l: 1000.000 1000.000 -2
1000.0 1000.0 -2
You can edit the above Fortran procedure to make it more general, e.g. to specify namelist file name and list name from the C program.