Trying to return array from Fortran to Matlab with mex, getting empty array instead

Question

So, I'm trying to return a an array of numbers from 1-n.

#include "fintrf.h"

C     Gateway routine
      subroutine mexFunction(nlhs, plhs, nrhs, prhs)

C     Declarations
      implicit none

C     mexFunction arguments:
      mwPointer plhs(*), prhs(*)
      integer nlhs, nrhs


      mwPointer mxGetPr
      mwPointer mxCreateDoubleMatrix
      mwPointer mxGetM, mxGetN
      mwPointer mrows, ncols
      mwSize size
      mwPointer x_ptr, y_ptr
      integer  x_input,i
      real*8, allocatable :: vec(:)
      x_ptr = mxGetPr(prhs(1))
      mrows = mxGetM(prhs(1))
      ncols = mxGetN(prhs(1))
      size = mrows*ncols

      x_ptr=mxGetPr(prhs(1))
      call mxCopyPtrToReal8(x_ptr,x_input,size)

      allocate (vec(x_input))

      do i=1,x_input
        vec(i)=i
      end do

      plhs(1) = mxCreateDoubleMatrix(1, x_input, 0)


      y_ptr = mxGetPr(plhs(1))
      call mxCopyReal8ToPtr(vec,y_ptr,x_input)



      deallocate ( vec )
      return
      end

I then call the mex file in fortran here

mex testingvec.F
Building with 'gfortran'.
MEX completed successfully.
a=testingvec(10);

and then find

a=[]

Can someone give me some help on this? If someone can give me some example code how to return a matrix as well, that would be sweet.

Thanks guys.

edit: new installment of the code. Still trying to get some help.

#include "fintrf.h"

C     Gateway routine
      subroutine mexFunction(nlhs, plhs, nrhs, prhs)

C     Declarations
      implicit none

C     mexFunction arguments:
      mwPointer plhs(*), prhs(*)
      integer nlhs, nrhs


      mwPointer mxGetPr
      mwPointer mxCreateDoubleMatrix
      mwPointer mxGetM, mxGetN
      mwPointer mrows, ncols
      mwSize size
      mwPointer x_ptr, y_ptr
      integer  i
      mwSize sizeone, x_input 
      integer*4 izero 
      real*8, allocatable :: vec(:)
      x_ptr = mxGetPr(prhs(1))
      mrows = mxGetM(prhs(1))
      ncols = mxGetN(prhs(1))
      size = mrows*ncols
      sizeone=1 
      izero=0 

      x_ptr=mxGetPr(prhs(1))
      call mxCopyPtrToReal8(x_ptr,x_input,size)

      allocate (vec(x_input))

      do i=1,x_input
        vec(i)=i
      end do

      plhs(1) = mxCreateDoubleMatrix(sizeone,x_input,izero)



      call mxCopyReal8ToPtr(vec,mxGetPr(plhs(1)),x_input)



      deallocate ( vec )
      return
      end

Answer 1

There were both declaration problems and issues with calls to mex functions. Here's a solution, which assumes that the input is an integer-valued double giving you the length of the output vector (assuming that this is what you wanted).

#include "fintrf.h"

C     Gateway routine
      subroutine mexFunction(nlhs, plhs, nrhs, prhs)

C     Declarations
      implicit none

C     mexFunction arguments:
      mwPointer plhs(*), prhs(*)
      integer*4 nlhs, nrhs


      mwPointer mxGetPr
      mwPointer mxCreateDoubleMatrix
      mwSize mxGetM, mxGetN
      mwSignedIndex mrows, ncols
      mwSize size, x_input, sizeone
      mwPointer x_ptr, y_ptr
      integer*4  i, izero, x_int
      real*8, allocatable :: vec(:)
      real*8 :: x_dbl

      sizeone = 1
      izero = 0



      !check input/output syntax
      if (nrhs /= 1) then
        call mexErrMsgIdAndTxt("MATLAB:testingvec:rhs",
    >            "Exactly 1 input variable required.")
      end if
      if (nlhs /= 1) then
        call mexErrMsgIdAndTxt("MATLAB:testingvec:lhs",
 >            "Exactly 1 output matrix required.")
      end if


      x_ptr = mxGetPr(prhs(1))
      mrows = mxGetM(prhs(1))
      ncols = mxGetN(prhs(1))
      size = mrows*ncols

      call mxCopyPtrToReal8(x_ptr,x_dbl,sizeone)
      x_input = int(x_dbl)

      allocate (vec(x_input))

      do i=1,x_input
        vec(i)=i
      end do

      plhs(1) = mxCreateDoubleMatrix(sizeone, x_input, izero)

      y_ptr = mxGetPr(plhs(1))
      call mxCopyReal8ToPtr(vec,y_ptr,x_input)



      deallocate ( vec )
      return
      end

I introduced a check for the number of input/output variables (to be updated in the actual program). And I introduced an auxiliary x_dbl which might or might not be necessary. This version reads the double input given to your function, and truncates it to get x_input.