How does MIR model discriminants?

[a1mer]

Here

MIR models discriminants more precisely. They should be thought of as a distinct field when it comes to borrows

How does MIR model discriminants? And what's the meaning of should be thought of as a distinct field?

Is it about enum?

Better having an example, thanks.

[alice]

They are modelled using the pattern known as a tagged union. You can find an example of that here, which I have copied in below:

// This Enum has the same representation as ...
#[repr(C)]
enum MyEnum {
    A(u32),
    B(f32, u64),
    C { x: u32, y: u8 },
    D,
 }

// ... this struct.
#[repr(C)]
struct MyEnumRepr {
    tag: MyEnumDiscriminant,
    payload: MyEnumFields,
}

// This is the discriminant enum.
#[repr(C)]
enum MyEnumDiscriminant { A, B, C, D }

// This is the variant union.
#[repr(C)]
union MyEnumFields {
    A: MyAFields,
    B: MyBFields,
    C: MyCFields,
    D: MyDFields,
}

#[repr(C)]
#[derive(Copy, Clone)]
struct MyAFields(u32);

#[repr(C)]
#[derive(Copy, Clone)]
struct MyBFields(f32, u64);

#[repr(C)]
#[derive(Copy, Clone)]
struct MyCFields { x: u32, y: u8 }

// This struct could be omitted (it is a zero-sized type)
#[repr(C)]
#[derive(Copy, Clone)]
struct MyDFields;

The above describes the layout that Rust guarantees for an #[repr(C)] enum, but it is also the layout that miri uses for #[repr(Rust)] enums (except that the *Fields structs are not marked #[repr(C)] for a #[repr(Rust)] enum.) Of course, this is not guaranteed behavior, since nothing is guaranteed about the layout of #[repr(Rust)] types.

[a1mer]

I mainly want to know about the meaning of the later sentence.

[alice]

Which sentence?

[Cerber-Ursi]

@a1mer

And what's the meaning of should be thought of as a distinct field ?

I think that means something like "if you only need to check the discriminant, MIR won't even try to borrow anything else".

[quinedot]

E.g.

let x = match some_option {
    None => 0,
    Some(_) => 1,
};

Only the discriminant need be examined.