Passing assumed shape to Fortran from JAVA JNA

[Chaitra_R_]

Hello,

My Fortran program contains the following:

REAL(KIND = R8), DIMENSION(:), ALLOCATABLE :: L, U
INTEGER :: N
N = 2
L(1:N) = (/0.0, 0.0/)

I modified the program and it looks like this now

INTERFACE
   SUBROUTINE inputs(L_array, arr_shape) BIND(C,name='inputs')
     IMPORT c_double,c_int,c_ptr
     type(c_ptr),value :: L_array
     integer(c_int), intent(in), dimension(2) :: arr_shape
   END SUBROUTINE inputs
END INTERFACE

SUBROUTINE inputs(L_array, arr_shape) BIND(C)
     USE ISO_C_BINDING
     USE VTdirect_MOD
     IMPLICIT NONE
       type(c_ptr),value :: L_array
       integer, intent(in), dimension(1) :: arr_shape_L
       real, dimension(:), pointer :: L
       integer :: n

       call c_f_pointer(L_array, L, arr_shape_L)
       print *, "L in fortran:", L
       do j = 1, n
            print *,L(j)
       end do
END SUBROUTINE

My JAVA program contains the following lines

double[] l = new double[]{0.0,0.0};
    int[] y = new int[]{2};

lib.inputs(l,x);

But, when I run this program,

I get the following output

L in fortran:   0.0000000000000000        0.0000000000000000

6.9518704107813286E-310
   1.3339772437713657E-322

What am I doing wrong?

[Steven_L_Intel1]

Since you declared the array "double" in Java, you should use real(C_DOUBLE) in the Fortran declaration of L.

[mecej4]

What am I doing wrong?

You posted code fragments that the compiler rejects, because there are a number of syntax errors. Obviously, the code that you posted is not the code that you ran. I cannot make sense of your claim that you modified a program such that it became a much bigger subroutine.

The interface block is of little use -- it is meant for use by callers of the subprogram concerned. Therefore, in Subroutine Inputs()  there are variables of undeclared type.

Please post the actual code. 

[FortranFan]

mecej4 wrote:

... I cannot make sense of your claim that you modified a program such that it became a much bigger subroutine.

..

Where is this claim?

[Steven_L_Intel1]

It's in the original post. Basically, a small Fortran program was turned into a Java program that called Fortran code. The Fortran code is larger than it needs to be.

[Chaitra_R_]

I looked into this post here - https://software.intel.com/en-us/forums/topic/282360 and I changed my SUBROUTINE inputs to look like this:

SUBROUTINE inputs(array_ptr, array_size) bind(c)
USE ISO_C_BINDING
IMPLICIT NONE

TYPE(C_PTR), VALUE :: array_ptr
INTEGER(C_INT), INTENT(IN), VALUE :: array_size
INTEGER(C_INT) :: i
REAL(C_DOUBLE), DIMENSION(:), POINTER :: array

call c_f_pointer(array_ptr, array, [array_size])

DO i =1, SIZE(array)
PRINT *, array(i)
END DO

END SUBROUTINE inputs

My Java program now contains the following lines of code:

double[] arr = new double[]{0.0,0.0};
int n = 2;
lib.inputs(arr,n);

The output is given below and as expected.

   0.0000000000000000     
   0.0000000000000000   

Thank you so much for all your help. Next time, I'll make it a point to include the full code and format my code properly.