Home > Software design >  How can I implement a default implementation of a generic trait for some generic type?
How can I implement a default implementation of a generic trait for some generic type?

Time:12-12

I want to implement a default implementation of a trait MovingAverage over some generic type Vec<T>, but I'm obviously doing something wrong since I can make it work for the concrete type Vec<f64> but not for the concrete type Vec<i32>. Here's the output from the error:

error[E0599]: the method `sma` exists for struct `Vec<{integer}>`, but its trait bounds were not satisfied
   --> src/main.rs:8:22
    |
8   |     let smai = numsi.sma(n);
    |                      ^^^ method cannot be called on `Vec<{integer}>` due to unsatisfied trait bounds
    |
   ::: /home/czar/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/alloc/src/vec/mod.rs:397:1
    |
397 | pub struct Vec<T, #[unstable(feature = "allocator_api", issue = "32838")] A: Allocator = Global> {
    | ------------------------------------------------------------------------------------------------ doesn't satisfy `Vec<{integer}>: MovingAverage`
    |
    = note: the following trait bounds were not satisfied:
            `f64: AddAssign<{integer}>`
            which is required by `Vec<{integer}>: MovingAverage`
            `f64: SubAssign<{integer}>`
            which is required by `Vec<{integer}>: MovingAverage`

Here's my code from ma.rs:

pub trait MovingAverage<Output = Vec<f64>> {
    fn sma(&self, periods: usize) -> Output;
}

impl<T> MovingAverage for Vec<T>
where
    T: Copy   Num,
    f64: AddAssign<T>   SubAssign<T>,
{
    fn sma(&self, periods: usize) -> Vec<f64> {
        let mut sum = 0f64;
        let mut ma = Vec::<f64>::new();
        for i in 0..self.len() {
            if i >= periods {
                ma.push(sum / periods as f64);
                sum -= self[i - periods];
            }
            sum  = self[i];
        }
        ma
    }
}

My main.rs

let numsf = vec![5., 10., 3., 9., 8., 7.];
let mut numsi = vec![2, 4, 3, 5, 1, 1];
let n = 2;
let smaf = numsf.sma(n);
let smai = numsi.sma(n); // doesnt work here 

What's going on in my error, and what would be the proper way to go about implementing a trait over a generic type without having to implement the trait for each concrete type? Thanks in advance, and kindly let me know if you need any further clarification.

CodePudding user response:

Maybe a better bound would be T: Into<f64>. Then you can simply convert each element into a f64 when you access it. Also, if you implement the trait for [T] instead of Vec<T>, it will also work for slices. Try this:

impl<T: Copy   Into<f64>> MovingAverage for [T] {
    fn sma(&self, periods: usize) -> Vec<f64> {
        let mut sum = 0f64;
        let mut ma = Vec::<f64>::new();
        for i in 0..self.len() {
            if i >= periods {
                ma.push(sum / periods as f64);
                sum -= self[i - periods].into();
            }
            sum  = self[i].into();
        }
        ma
    }
}

Playground

(Side note: I think your code may unintentionally ignore the last element of the input)

  • Related