Float::ldexp should use the same format as everything else

[mdinger]

std::num::Float::ldexp has an unusual format: it takes x: Self as it's first parameter where everything else in the trait takes self. It should take self as it's first parameter for consistency with everything else.

It is located in std::num::mod

[nagisa]

+1. There are a few different things to consider here, though:

1. ldexp does not operate on any FP number per se like most other operations do, but rather constructs a new FP number from two arguments one of which happens to be FP number. This function’s signature is consistent with most other methods that construct some kind of FP number (e.g. Float::zero()). We might want to consider whether methods such as next_after shouldn’t just go ahead and take x: Self instead.
2. ldexp can’t currently be chained, but I see literally 0 cases where anybody would chain this function.

cc @huonw

[mdinger]

Okay, but from a consistency standpoint, it's like 1/80 for x: self vs self. It it keeps the same format, it and it's similar functions would probably be better to go at the end or beginning of the function list where they won't be so unusual.

[aturon]

Nominating for 1.0-beta P-backcompat-libs

[pnkfelix]

One should be able to call this thing as a method, one way or another. We need to make sure that is resolved.

[pnkfelix]

1.0 beta, P-backcompat-libs

[milibopp]

I started to fix this.