problem implementing RC4/ciphersaber 2

[Spook]

After some meddling with my code and finally getting ciphersaber-1 to work, i thought it would be fairly simple to implement version 2. According the the webpage, one just has to add another loop around the key mixing phase. Excerpts follow :

i = j = 0; /* i and j are unsigned chars, as are all bytes in state[] and state2[] */
do {
	j = ( j + state[i] + state2[i] );
	x = state[i];
	state[i] = state[j];
	state[j] = x;
	i++;
} while (i);

The above works flawlessly on the test vectors on the site. Now, supposedly to use the improved version, one just needs to add a loop around they key mixing loop shown above.

/* n is a straight up integer */
i = j = 0;
while(n) {
	do {
		j = ( j + state[i] + state2[i] );
		x = state[i];
		state[i] = state[j];
		state[j] = x;
		i++;
	} while (i);
	n--;
}

However, the test vector, when decoded with the key "asdfg" and N=10 produces gibberish. Is there some flaw in the way I've set these loops up that I'm simply missing?

Ciphersaber Home page if this helps...

Thanks,
Spook

[redhead]

Your 'i' will only have a value of 0, the first time the outer loop is run, in the example on the page, it is initiated to 0 on each run.

/* n is a straight up integer */
j = 0;
while(n) {
      i = 0;
      do {
            j = ( j + state[i] + state2[i] );
            x = state[i];
            state[i] = state[j];
            state[j] = x;
            i++;
      } while (i);
      n--;
}

[Spook]

I tried your fix and got the exact same results in the output file; the condition while(i) at the bottom of the do loop only fails when i is equal to 0, so each pass of the inner loop ranges from 0..255 each time (remembering that i is an unsigned char).

[Valmont]

Give the rest of the code then.