f64::from(bool) and f32::from(bool)

[leonardo-m]

I think f64::from(bool) and f32::from(bool) could be added. Currently I perform that in two steps:

f64::from(u32::from(bool))
f32::from(u16::from(bool))

[hellow554]

Funny enough, the people are aware of this problem:

error[E0606]: casting `bool` as `f64` is invalid
 --> src/main.rs:3:20
  |
3 |     println!("{}", true as f64);
  |                    ^^^^^^^^^^^
  |
  = help: cast through an integer first

I think there's a valid reason for this, but let's see

[Mark-Simulacrum]

Integer casts are essentially no-ops since the bit pattern is the same, but casting to a float would be less obvious/straightforward, IMO.

I guess you expect 1.0 as the result, same as casting through an integer gets you?

[leonardo-m]

I expect f64::from(bool) to have about the same semantics of f64::from(u32::from(bool)), yes.

[workingjubilee]

Which zero to return?

[leonardo-m]

The same as f32::from(0_u8), that should be a +0.0.

[Kixunil]

Doc says it's implemented, so this should be closed?

But what the actual hell?! bool and a number are conceptually different, why allow any of those conversions? IMO the only sensible conversion is if x { 1 } else { 0 } (with compiler optimization). This is converting apples into eggs... Anyway, the ship has sailed.

[est31]

This should be closed now, yes. Implementation PR was #100390.