Field pattern shorthand can be interpreted as an item reference

[qnighy]

B { X } is a shorthand for B { X: X } and X here can be intepreted as an item reference, rather than a binding. I don't know if this is actually a bug or not. Possible fixes are:

• Leave it as is.
• Leave it as is with some warning.
• Force X be interpreted as a binding and let it fail to compile (due to local-global shadowing).

const X : i32 = 10;
struct B {
    X: i32,
}
fn main() {
    match (B { X: 9 }) {
        B { X } => println!("Foo!"),
        B { .. } => println!("Bar!"),
    }
}

• Current behavior: Bar! is printed.
• Possible expected behaviors: as-is or some warning/error emitted.
• Meta: rustc 1.20.0-nightly (ab5bec255 2017-06-22)

[Mark-Simulacrum]

Nominating for @rust-lang/lang. I don't think changing things here at this point is going to be easy (feature is stable) but perhaps we could do a warning cycle if we wanted to not allow this.

[petrochenkov]

I don't think this is a bug, just the shorthand desugaring and "existing entity vs fresh binding" disambiguation working orthogonally and as expected. (I.e. a perfect behavior for a corner case with near-zero practical impact.)

Force X be interpreted as a binding and let it fail to compile (due to local-global shadowing).

The wording for the "local-global shadowing" error is misleading, it's actually a "can't disambiguate between existing entity and fresh binding" error, not shadowing error (and it's reported more often than strictly necessary, I was planning to fix it, but still never got to this). If we can force X to be interpreted as a binding in some context, then we can disambiguate (in favor of the binding) and there should be no error.

[durka]

I don't see what this has to do with the shorthand? If the pattern is B { X: X } then it also prints "Bar!".

[eddyb]

Is there no warning on the field name of X (in the struct definition)?

[nikomatsakis]

I'm of the opinion that

(a) Bar { X } should always be equivalent Bar { X: X }
and (b) I would expect Bar { X } to be a binding

hence I'd prefer to have a lint targeting this case specifically.

[arielb1]

+1 for having this a lint

[steveklabnik]

Triage: not aware of any movement on this issue.