free a pointer to pointer variable

[sebas]

Hi, i'm trying to free a **pMatrix variable. (pointer to pointer). I have tried this code

for(i=0;i<=2*Nxfm+2*Nyfm;i++)
{
for(j=0;j<=2*Nxfm+2*Nyfm;j++)
{
free(pMatrix[i]);
}
}
free(pMatrix);

which is given in many turorials but this gives me an error (windows detect an iterruption point). why? my matrix is (2*Nxfm+2*Nxfm)x(2*Nxfm+2*Nxfm)=rows x columns.

Can anybody tell me where's the error? thanks in advance!

[sebas]

Sorry, the original code is

for(i=0;i<=2*Nxfm+2*Nyfm;i++)
{
free(pMatrix[i]);
}
free(pMatrix);

[iharrold]

This is C++ so you should use "new" and "delete" for dynamic memory allocation. Also use CODE tags.

new:

int ** pMatrix = NULL;

pMatrix = new int* [MAX1];
if (NULL != pMatrix)
{
   for (int i=0; i < MAX1; i++)
  {
      pMatrix[i] = new int[MAX2];
      if (NULL == pMatrix[i])
      {
          //allocation failed cleanup the previously allocated memory
      }
  }
}
else
{
    //Allocation failed
}

Delete:

if (NULL != pMatrix)
{
    for (int i = 0; i < MAX1; i++)
   {
        if (NULL != pMatrix[i])
            delete [] pMatrix[i];
   }

   delete [] pMatrix;
}

Or there abouts, didn't compile this. But you'll get the idea.

[sebas]

I think i dind't get the idea... this code fails. why you put the condition NULL != pMatrix?? What if I use calloc?
In my program the error is in the delete statement. Maybe is the compiler? I use visual c++ 2008.

[Disch]

- You can delete NULL pointers without any problem (nothing will happen). No need to check for NULL there.

- new doesn't return a NULL pointer like malloc does anyway, instead it throws an exception if allocation failed. The only time it returns a null pointer is if you use the nothrow version (which you're not using here)

- iharrold is right, if you're using C++ you should get in the habit of using new[]/delete[] instead of malloc/calloc/free. new/delete call ctors/dtor, malloc/free do not.

- nested new multidimensional arrays are evil. Obligatory link: http://cplusplus.com/forum/articles/17108/

- don't assume it's a compiler problem.

- it's difficult for us to show proper cleanup code without seeing the allocation code. Pretty much, the cleanup code should look exactly the same, but backwards.