Re: rand(), mt_rand() and limits

Hi! > Also, FWIW, anyone who used the Suhosin patch couldn’t use srand() > because it disabled it. And? You can disable any function in PHP, that doesn't mean anything. > Because if we don’t break it, people will continue to rely on it, and > this binds our hands for future versions. Sorry, that doesn't sound even remotely like a good reason. It doesn't provide any benefit to the user, but produces problems for those who use it. The gain for the users is zero, the added trouble - for those who use the feature - is substantial. > Also, those people will have their code break anyway if they upgrade > their OS and it changes its random number generator. If they used mt_rand that is not true. And OS RNG changes are not that frequent. E.g. the manual for srand on Mac (and, likely, all BSD) says: The srand() function sets its argument seed as the seed for a new sequence of pseudo-random numbers to be returned by rand(). These sequences are repeatable by calling srand() with the same seed value. The Linux manual says: The srand() function sets its argument as the seed for a new sequence of pseudo-random integers to be returned by rand(). These sequences are repeatable by calling srand() with the same seed value. I would say the expectation here is pretty clear. > Just because people do rely on it doesn’t mean they should or that we > should continue to allow them to. It doesn't mean they should, but it does mean it is a BC break, and one that produces no discernible gain at that. -- Stas Malyshev smalyshev@gmail.com