Confusing error message when same-named field and method both in scope

[brson]

/home/brian/dev/rust/src/test/run-pass/act.rs:178:18: 178:29 error: mismatched types: expected function or native function but found @mut bool
/home/brian/dev/rust/src/test/run-pass/act.rs:178               _ { msg.defer() }

I had both a field and a method called defer. rustc should say that both are defined and show me where.

[nikomatsakis]

Just as a side comment, this ambiguity would be addressed by using -> as the method operator.

[catamorphism]

Do we want to allow a class that has a field and a method with the same name? Or should the typechecker reject such a thing? (I re-read the spec and there's nothing about it.)

I would also be very cranky about using -> as the method operator (no good reason, I just think it's ugly), but I guess if we have to, we have to. Well, one dubious reason (it's also used for function return types, which doesn't really have anything to do with method calls).

[bstrie]

I naively figured that it would be illegal to have a field and a method with the same name. It seems like it would become a source of confusion. Niko mentioned on the mailing list that sometimes this is necessary to fulfill an iface, but I don't understand why that would be so.

[nikomatsakis]

Imagine an iface defined in some library X:

iface nameable {
    fn name() -> str;
}

and now a type defined in library Y:

type my_thing = {name: str};

and now I would like to define an impl so I can use my_thing as a nameable. I can't change the source of either one, as they come from different libraries I do not have control over.

[bstrie]

A link to the thread with Niko's proposal:

https://mail.mozilla.org/pipermail/rust-dev/2012-May/001819.html

I'm still a bit wary of using the same accessor notation for two distinct and potentially-colliding concepts. Perhaps \ could be used for method access? I don't think it's entirely unprecedented, but I can't recall where I've seen it before.

[graydon]

Possibly will get fixed by disambiguation between record field-selection and method-call. Niko will check back on this.

[catamorphism]

Blocked on @nikomatsakis landing AST changes.

[catamorphism]

Dup of #3446