For the C code in twice.c

int twice(int i)
{
return 2*i;
}

and Fortran code in cfunc.f90

program main
use iso_c_binding, only: c_int
implicit none
integer(c_int) :: i
interface
function twice(i) bind(c)
use, intrinsic :: iso_c_binding
integer(c_int) :: i
integer(c_int) :: twice
end function twice
end interface
i = 5
print*,i,twice(i)
end program main

I compile and run on Windows with

del *.o
if exist a.exe del a.exe
gcc -c twice.c
gfortran -c cfunc.f90
gfortran -o a.exe twice.o cfunc.o
if exist a.exe a.exe

but the output is

5 1182790712

What did I do wrong?

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Monday, February 21, 2022 at 6:47:46 PM UTC-5, Beliavsky wrote:

..
interface
function twice(i) bind(c)
use, intrinsic :: iso_c_binding
integer(c_int) :: i
integer(c_int) :: twice
end function twice
end interface
..

The VALUE attribute is missing in the dummy argument 'i' in the interface block.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Monday, February 21, 2022 at 8:28:36 PM UTC-5, FortranFan wrote:

On Monday, February 21, 2022 at 6:47:46 PM UTC-5, Beliavsky wrote:

..
interface
function twice(i) bind(c)
use, intrinsic :: iso_c_binding
integer(c_int) :: i
integer(c_int) :: twice
end function twice
end interface
..

The VALUE attribute is missing in the dummy argument 'i' in the interface block.

Thanks. As you say, it works when the declaration of i in the interface block is changed to

integer(c_int), value :: i

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 22/02/2022 01:47, Beliavsky wrote:

For the C code in twice.c

int twice(int i)
{
return 2*i;
}

In alternative, the argument in the C function could be a pointer:

int twice(int *i)
{
return 2**i; !In Fortran, it prints 2*i, not 2 ͥ
}

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

In my Fortran codes I specify the KIND of real variables but not integer variables. For compatibility with C it is
safer to declare integers as integer(c_int), but are there platforms where such an integer is distinct from the
default integer? In my code above if I replace integer(c_int) with integer and get rid of the "use iso_c_binding" statements,
it still works.

(Thanks to Ev. Drikos for his comment regarding pointers.)

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Beliavsky <beliavsky@aol.com> schrieb:

For the C code in twice.c

int twice(int i)
{
return 2*i;
}

and Fortran code in cfunc.f90

program main
use iso_c_binding, only: c_int
implicit none
integer(c_int) :: i
interface
function twice(i) bind(c)
use, intrinsic :: iso_c_binding
integer(c_int) :: i
integer(c_int) :: twice
end function twice
end interface
i = 5
print*,i,twice(i)
end program main

If you compile this with gfortran with -fc-prototypes,
you will get a valid C prototype for this function:

$ gfortran -fsyntax-only -fc-prototypes a.f90
#include <stddef.h>
#ifdef __cplusplus
#include <complex>
#define __GFORTRAN_FLOAT_COMPLEX std::complex<float>
#define __GFORTRAN_DOUBLE_COMPLEX std::complex<double>
#define __GFORTRAN_LONG_DOUBLE_COMPLEX std::complex<long double>
extern "C" {
#else
#define __GFORTRAN_FLOAT_COMPLEX float _Complex
#define __GFORTRAN_DOUBLE_COMPLEX double _Complex
#define __GFORTRAN_LONG_DOUBLE_COMPLEX long double _Complex
#endif

int twice (int *i);

#ifdef __cplusplus
}
#endif

Include this as a header in your C program, and your C compiler
will complain.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

"Beliavsky" wrote in message news:d5312537-6c84-4eaa-be4b-3d26d6253d2en@googlegroups.com...

In my Fortran codes I specify the KIND of real variables but not integer variables. For compatibility with C it is
safer to declare integers as integer(c_int), but are there platforms where such an integer is distinct from the
default integer? In my code above if I replace integer(c_int) with integer and get rid of the "use iso_c_binding" statements,
it still works.

With ifort you need the bind('C)' clause, but not in gfortran provided the names
match, perhaps via a *.DEF file. There is no guarantee that Fortran default integer is the same as C int, but it's pretty likely that it will be. One incompatibility that I have noticed is that Windows ultimately defines
DWORD as unsigned long which is 32 bits in MSVC++ but then cygwin defines unsigned long as 64 bits, so that messes up your Fortran module for Win32
API
interoperability.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hi,

On Feb 21, 2022, Beliavsky wrote
(in article<22b98749-9d21-43b5-8d59-596533b35fc4n@googlegroups.com>):

For the C code in twice.c

int twice(int i)
{
return 2*i;
}

and Fortran code in cfunc.f90

program main
use iso_c_binding, only: c_int
implicit none
integer(c_int) :: i
interface
function twice(i) bind(c)
use, intrinsic :: iso_c_binding
integer(c_int) :: i
integer(c_int) :: twice
end function twice
end interface
i = 5
print*,i,twice(i)
end program main

I compile and run on Windows with

del *.o
if exist a.exe del a.exe
gcc -c twice.c
gfortran -c cfunc.f90
gfortran -o a.exe twice.o cfunc.o
if exist a.exe a.exe

but the output is

5 1182790712

What did I do wrong?

My guess is, add value to the dummy argument declaration
in the interface block.

So --> integer(c_int) :: i
becomes --> integer(c_int), value :: i

Let me know how it goes . . .

--
Cheers!

Dan Nagle

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)