pre/post increment/decrement

[riemann]

Hi all guys, was browsing Linux kernel source and I came across this version of strstr:

char *strstr(const char *s1, const char *s2){
	size_t l1, l2;


	l2 = strlen(s2);
	if (!l2)
		return (char *)s1;
	l1 = strlen(s1);
	while (l1 >= l2) {
		l1--;
		if (!memcmp(s1, s2, l2))
			return (char *)s1;
		s1++;
	}
	return NULL;
}

in the

while

loop I was wondering if was possible to do instead something like:

while(l1-- >= l2)

and remove the line below the while expression..

So, it worked but is this correct? or are there any unexpected behaviour in doing a post-decrement inside a while?

thanks in advance

[grumpy]

Only if you want to change the behaviour of the function so it doesn't do what is intended.

[vart]

So, it worked but is this correct? or are there any unexpected behaviour in doing a post-decrement inside a while?

Just that original version is easier to read - so when somebody will need to modify it - less probability he will make a mistake due to misinterpreted code.

[riemann]

I didn't fully get your answers :/ despite it's less readable...does it present any issues releated with the post/pre increment/decrement operators? (like using them in a printf or things like that when there could be unexpected behaviours)

[rcgldr]

does it present any issues releated with the post/pre increment/decrement operators?

Since l1 is only used for the compare in while, it doesn't matter where the post decrement occurs. This should work also:

/* ... */
    for(l1 = strlen(s1); l1-- >= l2; s1++)
        if(!memcmp(s1, s2, l2))
            return (char *)s1;

or

/* ... */
    for(l1 = strlen(s1); l1 >= l2; l1--, s1++)
        if(!memcmp(s1, s2, l2))
            return (char *)s1;

[riemann]

Since l1 is only used for the compare in while, it doesn't matter where the post decrement occurs. This should work also:

/* ... */
    for(l1 = strlen(s1); l1-- >= l2; s1++)
        if(!memcmp(s1, s2, l2))
            return (char *)s1;

or

/* ... */
    for(l1 = strlen(s1); l1 >= l2; l1--, s1++)
        if(!memcmp(s1, s2, l2))
            return (char *)s1;

Great, gotcha now.. Thanks a lot!

[anduril462]

Yes, there is a difference in their behavior. They both perform the same number of iterations, but the value of l1 when the loop terminates will differ.

$ cat while.c
#include <stdio.h>

int main(void)
{
  int l1, l2, iter;


  l1 = 5;
  l2 = 2;
  iter = 0;
  printf("Starting decrement in loop body with l1 = %d, l2 = %d\n", l1, l2);
  while (l1 >= l2) {
    l1--;
    printf("iteration %d\n", ++iter);
  }
  printf("Finishing with l1 = %d, l2 = %d\n", l1, l2);


  l1 = 5;
  l2 = 2;
  iter = 0;
  printf("Starting decrement in loop condition with l1 = %d, l2 = %d\n", l1, l2);
  while (l1-- >= l2) {
    printf("iteration %d\n", ++iter);
  }
  printf("Finishing with l1 = %d, l2 = %d\n", l1, l2);


  return 0;

}
$ make while
gcc -Wall -ggdb3 -std=c99 -O0 -o while while.c -lm -lpthread -lrt
$ ./while
Starting decrement in loop body with l1 = 5, l2 = 2
iteration 1
iteration 2
iteration 3
iteration 4
Finishing with l1 = 1, l2 = 2

Starting decrement in loop condition with l1 = 5, l2 = 2
iteration 1
iteration 2
iteration 3
iteration 4
Finishing with l1 = 0, l2 = 2

With the decrement inside the loop body, l1 only decrements when the loop condition is true, meaning it doesn't decrement when l1 becomes small enough to cause the loop to stop. With the decrement in the loop condition, however, l1 decrements every time the loop condition is checked, which is every time the condition is true and when the condition is checked and fails. Thus, putting l1-- in the loop condition will leave l1 with a value 1 less than it would if the decrement is in the loop body.

Now, for that implementation of strstr that you referenced, it doesn't really matter, since the value of l1 is not used after the loop. But in general, such loops are not equivalent.

[rcgldr]

Since l1 is only used for the compare in while, it doesn't matter where the post decrement occurs.

Yes, there is a difference in their behavior. They both perform the same number of iterations, but the value of l1 when the loop terminates will differ.

Now, for that implementation of strstr that you referenced, it doesn't really matter, since the value of l1 is not used after the loop. But in general, such loops are not equivalent.

True. I didn't make it clear that if the value of l1 was being used after the loop exits that it would make a difference.