# Status of async fn in traits

A brief overview.

What we call "async fn in traits" is really multiple parts:

* Core support: ability to write `async fn` in traits + impls
* Bounding: ability to say `T: AsyncIter` but also "the future returned by `next()` is `Send`"
* Dynamic dispatch: ability to use `dyn AsyncIter` and invoke `async fn` methods
* RPITIT: sometimes you want an async fn that doesn't capture everything. This would most naturally be expressed with `-> impl Future` in the trait, though GATs + TAITs work too.

And some auxiliary/optonal parts:

* Refinements: ability to write `async fn` in trait but `-> impl Future` in impl, or other more specific future types, in impl

Current status:

| Part | Status |
| --- | --- |
| Core support | feature complete, but fixing bugs and writing tests |
| Bounding | have a proposal to address this that needs to be better circulated, ["return type notation"] |
| Dynamic dispatch | design space is enumerated, need to settle on a concrete design |
| RPITIT | feature complete, but needs cleanup and also authoring of tests; GATs + TAITs will allow this to be written in "desugared form"|

["return type notation"]: https://hackmd.io/9AH8Zr9ESMmur0n6Nix96w?view#Bounding-with-Send-Sync-and-other-bounds

Dynamic dispatch is a complex topic, but there are two main designs, and they come with their own dependencies:

| Design | Dependencies |
| --- | --- |
| Decide alloc strategy at call site ([part 9]) | returning of unsized values |
| Decide when creating dyn what adapter to use ([part 8]) | [`dyn*`] (started, needs elaboration) |

[part 9]: https://smallcultfollowing.com/babysteps/blog/2022/09/21/dyn-async-traits-part-9-callee-site-selection/
[part 8]: http://smallcultfollowing.com/babysteps/blog/2022/09/18/dyn-async-traits-part-8-the-soul-of-rust/
[`dyn*`]: http://smallcultfollowing.com/babysteps/blog/2022/03/29/dyn-can-we-make-dyn-sized/