RFC: Functional struct update for structs of different types but coinciding field names/types

[bblum]

I just thought of this and am not sure if there's some practical obstacle to it. Let's say you have:

struct Foo { a: T1, b: T2, c: T3 }
struct Bar { a: T2, b: T2 }

It would be nice to be able to write:

let bar: Bar = ...;
let foo: Foo = { c: type_3_value, ..bar };

[emberian]

I agree that this would be useful.

[huonw]

Triage: no change.

I'm personally ambivalent on this, but it might have some neat uses. FWIW, GHC is growing something tangentially related, "overloaded record fields", the core part of which is (in pseudo-Rust terms) essentially trait(s) HasField that give T: HasField<"foo", int> + HasField<"bar", f32> for

struct T {
    foo: int,
    bar: f32
}

so FSU would essentially become normal matching against HasField<"foo", T> for the fields foo: T that aren't explicitly listed. (This can't currently work directly for Rust, since we don't have strings in the type system.)

[emberian]

I'm going to close this, due to the new RFC process.