Understanding a trait thing

[jeramyRR]

Using the following code:

#[derive(PartialEq, Debug)]
pub struct USD(i32);
#[derive(PartialEq, Debug)]
pub struct GBP(i32);
#[derive(PartialEq, Debug)]
pub struct CAD(i32);

pub trait ToUSD {
    fn to_usd(&self) -> USD;
    fn convert<T: FromUSD>(&self) -> T {
        T::from_usd(&self.to_usd())
    }
}

impl ToUSD for GBP {
    fn to_usd(&self) -> USD {
        USD((self.0 * 130) / 100)
    }
}

pub trait FromUSD {
    fn from_usd(u: &USD) -> Self;
}

impl FromUSD for CAD {
    fn from_usd(u: &USD) -> Self {
        CAD((u.0 * 130) / 100)
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    #[test]
    fn it_works() {
        let g = GBP(200);
        let u = g.to_usd();
        assert_eq!(u, USD(260));

        let c = CAD::from_usd(&u);
        assert_eq!(c, CAD(338));

        let c2: CAD = g.convert();
        assert_eq!(c2, c);
    }
}

I don’t understand why the last assert_eq! passes. I did not implement FromUSD for GBP. What am I missing here?

[OptimisticPeach]

fn convert<T: FromUSD>(&self) -> T {
    T::from_usd(&self.to_usd())
}

✓ CAD: FromUSD
✓ Self = GBP: let g = GBP(200)
✓ Self: ToUSD

Therefore, convert is auto implemented for GBP anywhere where T: FromUSD, and therefore this function call works.

I did not implement FromUSD for GBP.

You don’t need to impl FromUSD for GBP, because it is a constraint on T, not Self, and T is CAD.

[jeramyRR]

Thanks, I get it now.