Suggestion to make this a trait

[akberg]

Hi!
I found this crate for the same reason you created it – I find myself using decimal rounding so often it bugs me that there is no built-in support for it, so I would check if a crate for it existed before going on and creating one myself. But, for this API to flow better with the Rust APIs, I do have some suggestions:

• Use f64::powi/f32::powi to remove the limitation set by hard-coding each order of magnitude (saw that this is exactly what more freedom to round #1 suggested 3 years ago, so I'll consider this crate dead): Sorry, this was already added, I was looking at an outdated commit.

fn round(n: f32, i: u32) -> f32 {
  (n * 10u32.pow(i) as f32).round() / 10u32.pow(i) as f32
}

• The std already has a function called round (f32::round() -> u32), consider renaming it to e.g. roundf to emphasize that the return value is a float.
• Consider making it a trait instead (or in addition):

trait RoundF {
    fn roundf(&self, rounding: i32) -> Self;
    fn ceilf(&self, rounding: i32) -> Self;
    fn floorf(&self, rounding: i32) -> Self;
}
impl RoundF for f32 {
    fn roundf(&self, rounding: i32) -> f32 {
        (self * 10.0f32.powi(rounding)).round() / 10.0f32.powi(rounding)
    }
    . . .
}
impl RoundF for f64 {
    fn roundf(&self, rounding: i32) -> f64 {
        (self * 10.0f64.powi(rounding)).round() / 10.0f64.powi(rounding)
    }
    . . .
}