What is the difference between `e1` and `&e2` when

问题:

The compiler says that e1: &i32 and e2: i32. Having read the docs for slice::Iter and the book chapter on loops, I'm still confused.

More generally, can a particular element in a slice be owned? It seems like in case 2, e2 is owning an element, is it?

fn main() {
    let slice = &[1, 2, 3];
    for e1 in slice.iter() {
        println!("{}", e1); // case 1
    }

    for &e2 in slice.iter() {
        println!("{}", e2); // case 2
    }
}

回答1:

When used in a destructuring pattern binding, the ampersand & is used to dereference a value:

let a_number: i32 = 42;
let a_reference_to_a_number: &i32 = &a_number;

let another_reference = a_reference_to_a_number;
let &another_number = a_reference_to_a_number;

let () = another_reference; // expected type `&i32`
let () = another_number;    // expected type `i32`

This applies anywhere a pattern is accepted, such as in a let or if let, as a function argument, the for-loop variable, or a match arm.

---

While initially confusing to many people, this is actually the language being consistent with how enums and structs are pattern matched and thus removed from the inner variable binding:

let a_value: Option<&i32> = Some(&42);

if let Some(&val) = a_value {
    let () = val; // expected type `i32`
}

Note how val is no longer "wrapped" by the Some, just like it is no longer "wrapped" by the reference.

This behavior of & in patterns is why the ref keyword is needed. The ref keyword is used to unambiguously introduce a reference in a pattern match.