Pre-RFC: Rank-N universal quantification (functions / closures)

[hadronized]

I had opened a link on Reddit (here) about rank-N universal quantification but I didn’t get that much of a feedback, so I guess the Pre-RFC will have more inertia here.

Rendered

[glaebhoerl]

@phaazon The right place for pre-RFCs is internals.rust-lang.org.

[hadronized]

@glaebhoerl define right, please. Because last time I posted there, no one ever answered.

[Centril]

@phaazon I wrote a draft some time ago about the same idea and more but I've been busy with some other stuff... There's interest from some language team members (including me) re. this idea, but we should sync our work together.

We might want to approach this in a staged manner:

1. Introduce the quantified for<..> |$pat, ..| $expr closure syntax.
2. Introduce for<'a: 'b> as a syntax.
3. Permit quantification on for<T: Debug> and for<const A: B>
4. Permit type/const generic closures

cc @cramertj @nikomatsakis

[hadronized]

Interesting! What would (1.) do?

[Centril]

@phaazon You'd be able to explicitly denote lifetime-generic closures, e.g for<'a> |x: &'a u8| x. We could introduce everything at once, but incremental progress might be easier to deal with. :)

[cramertj]

yup, this is something we clearly want and the design is pretty obvious IMO but actually doing here is blocked on chalk (AFAIK)

[hadronized]

Alright. I’ll add more teaching section to the RFC. Feel free to open a PR with the lifetime and |_| notation!