std::rand::{gen_int_range,gen_uint_range}

[sruggier]

The gen_uint_range function in std::rand uses the modulo operator to remap generated numbers from [0, uint::max_value] into the range [start, end). If uint::max_value is not a multiple of (end - start), then this technique will result in random numbers of an uneven probability distribution. For small values of (end - start), the error is negligible, but as the size of (end - start) approaches the size of uint::max_value, there starts to be a non-negligible error.

To illustrate the problem, imagine if (start, end) were (0, uint::max_value/3*2). All the random numbers above 2/3 of the max_value would be remapped to the range of [0, uint::max_value/3 ], and it would be twice as probable that the final number is within the first half of the requested range.

[huonw]

I have a fix. (It also merges the two functions and makes them more generic.)