Re: Re: qsort fix

On Fri, 28 May 1999, Rasmus Lerdorf wrote: > > I actually think it's legitimate. If your comparison function isn't > > deterministic, the sorting algorithm may not make sense, and it > > assumptions one makes when designing a sorting algorithm (at the very > > least, that there's an order defined on the to-be-sorted set) end up not > > being valid, and may screw the algorithm. > > Then that should be stated. Especially when all other implementations > don't crash on this. It actually may be luck rather than anything else. When your code screws up it can end up undetected, and it can end up crashing. I think that when a sorting function asks for a comparison function that compares two elements, and should return 0 if they're equal, and a descriptive number if they're not - it's pretty obvious it means it should be deterministic. That's what comparison functions are, much like you wouldn't expect operator< to return random information on you either if your operands are identical. Zeev -- ----------------------------------------------------- Zeev Suraski <zeev@zend.com> For a PGP public key, finger bourbon@netvision.net.il -- PHP Development Mailing List http://www.php.net/ To unsubscribe send an empty message to php-dev-unsubscribe@lists.php.net For help: php-dev-help@lists.php.net