Revised UFCS proposal

[nikomatsakis]

A proposal to add an unambiguous syntax for referring to methods and other associated items. An adapted version of the original proposal by @nick29581.

---

currently active: https://github.com/rust-lang/rfcs/blob/master/text/0132-ufcs.md

[mstewartgallus]

How does this relate to function pointers and closures? Will we be able to do something like x.map(Trait::method) like Java 8?

[Ericson2314]

Could trait methods be given an extra type parameter corresponding to the type of the implementer when bound as a function?

e.g.

trait Foo {
    fn method<A: Bar>(&self, bob: &Self) -> Box<A>;
}

binds

pseudo-mod Foo {
    fn method<T: Foo, A: Bar>(alice: &T, bob: &T) -> Box<A>
}

Edit: to disambiguate between trait object member and trait method, I supposed one can call it with explicit type/lifetime parameters (<_>, vs <_,_> in this case). If trait object methods cannot be used in a higher order context that should work. But I wouldn't mind this and something different to unambiguously refer to the trait object type itself, the <_,...> trick isn't the most pleasant.

[telotortium]

Typo to_struivalent -- you ran :%s/eq/to_str/g?

[jsanders]

Yeah I think he was using Eq for all the examples and switched to ToStr - gave me a little chuckle.

[bill-myers]

I think the most natural syntax for <T as ToStr>::to_str would be (impl ToStr for T)::to_str, but unfortunately it's longer.