Hi John,

You asked:

JM>I can't program in C, I prefer various types of asm, but I have
JM>had a go at writing a C program to generate prime numbers...

JM>#include "stdio.h"
JM>#define m 100
JM>main() {int i=1,j,p=1,fl,l[m]; while (pwhile (jif (fl) {l[p++]=i; printf("%d\n", i);} } }

JM>Okay, here's the problem :-) - it compiles okay using a couple of
JM>freeware compilers, and prints out the primes nicely. However, when
JM>I compiled it on a Vax, using a 'real' compiler,it just printed out
JM>a list of odd numbers...

Typical...

JM>I can't see what the problem is. I have a suspicion it is something
JM>to do with the handling of FL. Can anyone help me here...

No, it isn't really. It's an initialisation problem with p and l[]. You
initialise p to 1, so there is never any value stored in l[0], but
the first time through for a number (with j = 0) you test for
termination of the while on "l[0] * l[0] <= i". If l[0] has a random
value left in it (as it commonly will) you will fail on the first while
without ever reaching the testing for dividors, so all tested values
will be printed.

My suggested changes (with comments).
I expanded it to make it more readable.

#include "stdio.h"

#define m 100

/* For ANSI C I prefer to be explicit and use */
int main (void)
{
unsigned int i = 1,j,p = 0,fl,l[m] = {3,5};
/* Initialise p = 0, not 1 */
/* Initialise l[] with the first known primes expected */
while (p < m)
  {
  i += 2;
  fl = 1;
  j = 0;
/* You had this test. */
/*  while (j < p && fl && l[j] * l[j] <= i) */
/* By using a break when fl is cleared you can simplify the test to */
  while (j < p && l[j] * l[j] <= i)
/* which should be quicker */
    if (i % l[j++] == 0)
      {
      fl = 0;
      break;
      }
  if (fl)
    {
    l[p++] = i;
    printf ("%u\n", i);
    }
  }
/* We need a return value from main() */
return 0;
}

George

 * SLMR 2.1a * Computers eliminate spare time.

--- Maximus/2 3.01