std::num::pow: exponent should not be a uint

[vks]

Right now pow has the following signature:

pub fn pow<T: One + Mul<T, T>>(base: T, exp: uint) -> T

It does not really make sense that the exponent is a pointer-sized uint. This makes this function unnecessarily platform-dependent.
Also, it is awkward to use with BigInt as an exponent. (This mainly an ergonomic issue, as a uint should be large enough to store any reasonable exponent.)

I think it would be better to have a signature like

pub fn pow<T: One + Mul<T, T>, U: One + Zero + Add<U, U> + Div<U,U>>(base: T, exp: U) -> T

(Note that pow currently uses some bitwise operations for efficiency, the proposed type constraints do not reflect that.)

Alternatively, to keep it simple the exponent could just be a constant-size unsigned integer (u32 should be sufficient [1]).

[1] pow(1.1, pow(2, 32)) ~= pow(10, 177780229)

[gsingh93]

This is a valid issue that I want fixed, but it's in the wrong repo now. Could someone moves this over to https://github.com/rust-lang/num?