Random numbers

[Anon]

Hi, I need a script that will make a random number that nobody else will be able to guess...I tried writing the following script, but it comes up with the same number twice every now and then, and that can not happen. Is anything wrong with it?

$random1 = rand(0,99);
$random2 = rand(0,99);
$random3 = rand(0,99);
$random4 = rand(0,99);
$random5 = rand(0,99);
$random6 = rand(0,99);
$random7 = rand(0,99);
$random8 = rand(0,99);
$random9 = rand(0,99);
$random10 = rand(0,99);
$random11 = rand(0,99);
$random12 = rand(0,99);
$random13 = rand(0,99);
$random14 = rand(0,99);
$random15 = rand(0,99);

$randomnum = $random1 . $random2 . $random3 . $random4 . $random5 . $random6 . $random7 . $random8 . $random9 . $random10 . $random11 . $random12 . $random13 . $random14 . $random15;

echo $randomnum;

[lawnmowerman]

Check out the manual page for srand():
http://www.phpbuilder.com/manual/function.srand.php

srand() seeds the rand() function to return (hopefully) unique data as long as the seed is somewhat unique.

The example on the srand() page shows how to use the microtime() function to seed rand().

I generally use the time() function, instead, since every second the seed would change.

Hope this helps!

-Rich

[polson]

Yes, you need to plant the seed (as described within rand() manual entry), the example under srand() in manual works nicely. And remember, just one call of srand per page no matter how many times you call rand, just call srand once.

And yes, the same number may or may not come up twice as it's random. If limitations were setup to disallow numbers from coming up twice, they then would not be random. PHP doesn't remember what numbers previous renditions of rand() created.

[kparker]

What Phil and Rich said, except I disagree about using time() instead of microtime()--if your site is busy (hope, hope!) you can expect to have peak times when quite a few visitors might hit the same page within the same 1-second interval.

A further recommendation is to use mt_rand() instead of plain rand(). mt_rand() is both numerically better and also faster, so it's a real win-win.

[lawnmowerman]

Ah...my mistake. After re-reading the microtime() page, I agree that microtime() is better than time().

I was under the impression that microtime() only returned the number of msecs since the last second, and not the rest of the timestamp...though this is how the example works. In a mission critical application, I'd probably use the entire output of microtime() and use both the UNIX timestamp AND microseconds to make a seed.

....but I digress 🙂

-Rich