Reply to thread

Message

[QUOTE="Michael Mair, post: 2468590"]
It looks right -- had you formatted the code in a more
readable manner, this would have been easier.
Three remarks:
1) You are "admitting" that your implementation is based on
  rand().
  I'd rather use an
   void init_rand_in_range (long seed);
  function from the start -- even if it only wraps srand(), this
  will make it easier if you ever want to use another pseudorandom
  number generator -- just "compile and link" instead of "replace
  all srand() calls in every bit of code by an init....() call".
2) In older rand() implementations, there often were problems
  with the modulo approach as you got shorter cycle lengths
  than for the division approach (see below).
3) Be verbose:
int rand_in_range(int m, int n)
{
   /*seed srand in main */
   /* [m, n] is range */
   int roll_again_threshold, divisor, result;
   int offset = m;
   int num_results = n - m + 1;

if (num_results == 1) {
     return m;
   }
   if (num_results <= 0) {
     num_results = m - n + 1;
     offset = n;
   }

roll_again_threshold = RAND_MAX - RAND_MAX%num_results;
   divisor = roll_again_threshold/num_results;

do {
     result = rand();
   } while (result >= roll_again_threshold);
   result /= divisor;
   return offset + result;
}

This is exaggerated but IMO more helpful than your code:
Why say "diff + 1" if you mean the number of results?
Or t and p? top vs roll_again_threshold is mostly a matter
of taste -- the latter offers itself for the division.

Cheers
  Michael
[/QUOTE]

Verification