Tracking issue for Captures trait

[alexreg]

This solution was initially suggested by @nikomatsakis in this comment.

Steps:

• Implement the feature.
• Adjust documentation.
• Stabilize feature.

Questions:

• Is this a permanent solution?
• Does this need a FCP?
• Can we get a minimal use case for this feature?

CC @alexcrichton

[alexreg]

Maybe someone in @rust-lang/libs could kindly add the appropriate tags?

[ExpHP]

Github is making the comments a nuisance to navigate. Is there a cliff-notes version of how this is different from impl Trait + 'a, and the problem it solves?

(I imagine that, since it's a trait bound, it introduces invariance over 'a as opposed to covariance?)

[alexreg]

@ExpHP I still don't 100% Niko's comment, but I think you're roughly right at least. It's definitely related to invariance somehow. Here is the comment, reproduced.

So #49041 contains a fix for #46541, but that fix has more impact than I anticipated -- e.g., the compiler doesn't bootstrap now -- and it's giving me a measure of pause about the right course here. The problem in #49041 is that we could accidentally allow lifetimes to leak out that we were not supposed to. Here is how this manifests in the compiler. We might have a method like this:

impl TyCtxt<'cx, 'gcx, 'tcx>
where 'gcx: 'tcx, 'tcx: 'cx
{
    fn foos(self) -> impl Iterator<Item = &'tcx Foo> + 'cx {
        /* returns some type `Baz<'cx, 'gcx, 'tcx>` that captures self */
    }
}

The key thing here is that TyCtxt is invariant w/r/t 'tcx and 'gcx, so they must appear in the return type. And yet only 'cx and 'tcx appear in the impl trait bounds, so only those two lifetimes are supposed to be "captured". The old compiler was accepting this because 'gcx: 'cx, but that's not really correct if you think about the desugaring we have in mind. That desugaring would create an abstract type like this:

abstract type Foos<'cx, 'tcx>: Iterator<Item = &'tcx Foo> + 'cx;

and yet the value for this abstract type would be Baz<'cx, 'gcx, 'tcx> -- but 'gcx is not in scope!

The workaround here is that we have to name 'gcx in the bounds. This is kind of annoying to do; we can't use 'cx + 'gcx. We can I suppose make a dummy trait:

trait Captures<'a> { }
impl<T: ?Sized> Captures<'a> for T { }

and then return something like this impl Iterator<Item = &'tcx Foo> + Captures<'gcx> + Captures<'cx>.

[oli-obk]

We do have this trait in

rust/src/librustc/util/captures.rs

Line 6 in 9d9c7ad

pub trait Captures<'a> { }

If we move it to core::marker (unstably) we can use it in the printing of async fn return types, which currently just emit nothing:

rust/src/librustc/hir/print.rs

Line 368 in 9d9c7ad

hir::TyKind::Def(..) => {},

So with Captures available in core, an async fn foo<'a>() -> Bar could emit something like fn foo<'a>() -> impl SomeGeneratorThingy<Bar> + Captures<'a>

[SimonSapin]

Since this trait is defined in librustc, it doesn’t appear to be part of the standard library at all. I’ve changed the team labels accordingly, please revert if this is in error.

Or is the plan to move the trait to core::marker? Why is this a tracking issue in the first place?

impl Trait in return type position is now stable. Is it unsound when this kind of lifetimes are involved?

[alexreg]

@SimonSapin Thanks. However, this issue is really superseded by #60670.

And no, there is no unsoundness. The point is it is overly-restrictive at present.