I know there are many programming experts here in Sajha, I am going to put here a simple problem I faced today seeking your suggestions.

I am trying to write a numeric array into a binary file using fwrite. The variable is three dimensional converted into two dimensional double type, var[k][t], where [k] contains two-dimension inside. k is not sequential but has to be picked up from i and j released by separate routine. I coded

fwrite(var[k][t], sizeof(var[0][t]), sizeof(var[k][t]), FILE*);

but it displayed the following error message,

error: cannot convert `double' to `const void*' for argument `1' to `size_t fwrite(const void*, size_t, size_t, FILE*)'

I have seen fwrite is used successfully in writing character arrays. How can I use this to write numeric array?

It seems that I need to add something to successfully convert 'double' to 'const void*', but did not get any clue on that.

Any help or suggestions are welcome.

!!!

cast the return of the double.. it might work or barf for some other reason later.. but u will b able to pass this one..

(const void * (var [k][t]) )

this should do it.. for ur compilation

From the error message and your code fragment it seems that var is declared as a 2 dimensional array (look at the error message, it says cannot convert 'double' instead of 'double *' => var[k][t] is double instead of double *).

The first argument to fwrite() has to be a pointer. You can use type casting but make sure that the pointer points to contiguous block of memory (might not be contiguous if you dynamically allocate 2 or higher dimensional arrays).

The third argument to fwrite() in your code fragment seems to have some problems. Although it is correct syntactically, it might have some semantic issues. sizeof(var[k][t]) is sizeof(double) if var is 2D or sizeof(double *) if var is 3D. The point is that it will always be constant (unless this is what you want).

Thanks Thapap and |<1$|-| for valuable suggestions.

Thapap's idea worked out to solve the compiling error message when I added
const void * (var [k][t]) ;
just before the fwrite statement. But it produced segmentation fault while running the program.

It is true the var[k][t] is declared as two dimensional double. The memory allocation is dynamic. Declaring const void*(var[k][t]) in the beginning of program caused further problem of conflict in declaration, with the following error message.

error: conflicting declaration 'const void*var[((long unsigned int)((long int)k))][((long unsigned int)((long int)t))]'

All examples I could find on fwrite illustrate just writing character type data array. Could not find any good referece to write the numeric data array in binary form,

Here is a sample code to write a 2D array of double to a file and read it back. Things work similarly for higher dimensions and dynamically assigned arrays.

#include
#include

int main()
{
FILE * opfile, * readfile;
double var[5][5], varr[5][5];
int i,j,k;

for(i = 0; i < 5; i++)
for(j = 0; j < 5; j++)
var[i][j] = i+j+1;

opfile = fopen("storefile", "w");
if(opfile == NULL){
printf("Oops!\n");
return 0;
}

fwrite(var, sizeof(double), 25, opfile);

fclose(opfile);
readfile = fopen("storefile", "r");

fread(varr, sizeof(double), 25, opfile);
for(i = 0; i < 5; i++){
for(j = 0; j < 5; j ++)
printf("%f %f\n", varr[i][j], var[i][j]);
}

return 0;
}

View/Share this post only

heres the code again:

 
 #include <stdio.h> 
 #include <stdlib.h> 
  
 int main() 
 { 
  FILE * opfile, * readfile; 
  double var[5][5], varr[5][5]; 
  int i,j,k; 
  
  for(i = 0; i < 5; i++) 
  for(j = 0; j < 5; j++) 
  var[i][j] = i+j+1; 
  
  opfile = fopen("storefile", "w"); 
  if(opfile == NULL){ 
  printf("You suck\n"); 
  return 0; 
  } 
  fwrite(var, sizeof(double), 25, opfile); 
  
  fclose(opfile); 
  readfile = fopen("storefile", "r"); 
  
  fread(varr, sizeof(double), 25, opfile); 
  for(i = 0; i < 5; i++){ 
  for(j = 0; j < 5; j ++) 
  printf("%f %f\n", varr[i][j], var[i][j]); 
  } 
  
  return 0; 
 } 
  
 

looks like ^ bro overtook me hehe.

but since i put my 5 mins into it, here's how i would write numeric values to a binary file :P
========

#include
#define size 3
int main ()
{

FILE * pFile;
struct var{
double x1;
double x2;
};
typedef struct var var;

var vararr[size]={{3.4,4.5},{5.4,6.1},{7.1,8.9}};

pFile = fopen ( "file.bin" , "wb" );
fwrite (vararr , sizeof(var) , size , pFile );

fclose (pFile);
return 0;
}
======

View/Share this post only

Thanks |<1$|-| and Loote for sample codes. Sajha members rock.