What differences between win32 and win64 might cause error in executing mex file?

Question

I am trying to run example provided by field II on my computer. This example contains matlab functions as well as mex32 file. I have later version of matlab so i take initial .c file and converted it into mex64. However despite setting everything like in provided example my mex function does not work. I am not that familiar with c programming so i am not sure where exacly in the main function problem occurs. If anyone please could direct me what is going wrong in this .c program. Also, can i succesfully transform this c code to matlab function instead of using mex file or is there some nuances in the C code that can not be fully captured in the MATLAB translation?

Matlab error full original C code

I was trying to write matlab function based on algorithm from C program but i am not sure if i am getting indexing correctly. I am aware that matlab indices starts from 1 not 0. What i don't understand is this env_pointer being set to be one variable's address and then is called like it contains N_samples+1 variables.

void make_interpolation (unsigned char  *envelope_data,   /*  The envelope detected and log-compressed data */
                 int            N_samples,        /*  Number of samples in one envelope line        */
             
                 int            *index_samp_line, /*  Index for the data sample number              */
                 int            *image_index,     /*  Index for the image matrix                    */
                         float          *weight_coef,     /*  The weight table                              */
                         int            N_values,         /*  Number of values to calculate in the image      */
             
                         unsigned char *image)            /*  The resulting image                           */

{int           i;                 /*  Integer loop counter                */
 int           ij_index_coef;     /*  Index into coefficient array        */
 unsigned char *env_pointer;      /*  Pointer to the envelope data        */
 float         *weight_pointer;   /*  Pointer to the weight coefficients  */
 
  ij_index_coef = 0;
  for (i=0; i<N_values; i++)
    {
     weight_pointer = &(weight_coef[ij_index_coef]);
     env_pointer = &(envelope_data[index_samp_line[i]]);
     image[image_index[i]] 
              =         weight_pointer[0] * env_pointer[0]
                      + weight_pointer[1] * env_pointer[1]
                      + weight_pointer[2] * env_pointer[N_samples]
                      + weight_pointer[3] * env_pointer[N_samples+1] + 0.5;
     ij_index_coef = ij_index_coef + 4;
    }
}

Answer 1

At first glance this is the obvious problem in the original source code: The dimensions array needs to be the proper integer size for API calls, so you need to change this:

 int dimension[2];  

to this

 mwSize dimension[2];   

Also, you may get a compiler warning on this conversion:

(unsigned char *) mxGetPr(etc)

so I would change it to this instead:

(unsigned char *) mxGetData(etc)