09 May 2023

Giving, lending, and async closures

In a previous post on async closures, I concluded that the best way to support async closures was with an async trait combinator. I’ve had a few conversations since the post and I want to share some additional thoughts. In particular, this post dives into what it would take to make async functions matchable with a type like impl FnMut() -> impl Future<Output = bool>. This takes us down some interesting roads, in particular the distinction between giving and lending traits; it turns out that the closure traits specifically are a bit of a special case in turns of what we can do backwards compatibly, due to their special syntax. on!

(read more...)

---

03 April 2023

Fix my blog, please

It’s well known that my blog has some issues. The category links don’t work. It renders oddly on mobile. And maybe Safari, too? The Rust snippets are not colored. The RSS feed is apparently not advertised properly in the metadata. It’s published via a makefile instead of some hot-rod CI/CD script, and it uses jekyll instead of whatever the new hotness is.¹ Being a programmer, you’d think I could fix this, but I am intimidated by HTML, CSS, and Github Actions. Hence this call for help: I’d like to hire someone to “tune up” the blog, a combination of fixing the underlying setup and also the visual layout. This post will be a rough set of things I have in mind, but I’m open to suggestions. If you think you’d be up for the job, read on.

(read more...)

---

29 March 2023

Thoughts on async closures

I’ve been thinking about async closures and how they could work once we have static async fn in trait. Somewhat surprisingly to me, I found that async closures are a strong example for where async transformers could be an important tool. Let’s dive in! We’re going to start with the problem, then show why modeling async closures as “closures that return futures” would require some deep lifetime magic, and finally circle back to how async transformers can make all this “just work” in a surprisingly natural way.

(read more...)

---

16 March 2023

Must move types

Rust has lots of mechanisms that prevent you from doing something bad. But, right now, it has NO mechanisms that force you to do something good¹. I’ve been thinking lately about what it would mean to add “must move” types to the language. This is an idea that I’ve long resisted, because it represents a fundamental increase to complexity. But lately I’m seeing more and more problems that it would help to address, so I wanted to try and think what it might look like, so we can better decide if it’s a good idea.

(read more...)

---

15 March 2023

Temporary lifetimes

In today’s lang team design meeting, we reviewed a doc I wrote about temporary lifetimes in Rust. The current rules were established in a blog post I wrote in 2014. Almost a decade later, we’ve seen that they have some rough edges, and in particular can be a common source of bugs for people. The Rust 2024 Edition gives us a chance to address some of those rough edges. This blog post is a copy of the document that the lang team reviewed. It’s not a proposal, but it covers some of what works well and what doesn’t, and includes a few sketchy ideas towards what we could do better.

(read more...)

---

12 March 2023

To async trait or just to trait

One interesting question about async fn in traits is whether or not we should label the trait itself as async. Until recently, I didn’t see any need for that. But as we discussed the question of how to enable “maybe async” code, we realized that there would be some advantages to distinguishing “async traits” (which could contain async functions) from sync traits (which could not). However, as I’ve thought about the idea more, I’m more and more of the mind that we should not take this step — at least not now. I wanted to write a blog post diving into the considerations as I see them now.

(read more...)

---