carl.tokio

dial9: a flight recorder for Tokio. It collects the Tokio runtime metrics and renders them in a useful way to debug performance issues. tokio.rs/blog/2026-03-1…

We're setting aside 20 free TokioConf tickets for contributors and open source maintainers. If you've submitted a PR to any tokio-rs, hyperium, or tower-rs repo, that's you. Know someone who qualifies? Send them our way! tokio.rs/blog/2026-03-1…

finished today the slides for my talk for TokioConf "I want it, you want it, we want it: Let's rewrite it in Rust!" Slides coming up great and hoping this meme will resonate. Come to the conf in Portland, Apr 21st!

Announcement time 📢 We’re excited to welcome Carl Lerche, author of Tokio, to our next RustRover livestream. @carllerche and @vbragilevsky will talk about the journey of async Rust, what shaped it, and how it became a core part of modern Rust applications. 📆 Feb 12, 17:00 UTC 👉 Save your spot: info.jetbrains.com/rustrover-live…

On episode 6 of High Leverage, Joe Ruscio (@josephruscio) joins Carl Lerche (@carllerche), the creator of Tokio (@tokio_rs), for a deep technical discussion on async programming and infrastructure design. Tune in! hubs.ly/Q040QsRg0

@BansodMahesh @fasterthanlime Someone else already replied, but to repeat, you don't have to run the cli tool by hand. Add it to build.rs (will have an example soon) and it will be handled automatically for you.

@fasterthanlime m curious why you think so. I'll have to manually run the cli tool to generate code when something changes, while proc macro would just build the right thing

@ConradLudgate PR wen? We should discuss more before doing more work. Seriously good job.

With 1 line of code added, you can get a nice breakdown automatically of where the runtime is getting blocked. All it takes is adding `Supervisor::new(&tokio::runtime::Handle::current()).spawn(Duration::from_millis(100));` at the start of async fn main().

My friends, it's finally time to talk about System Initiative. You can learn all about it at systeminit.com, including a demo video with cats. youtube.com/watch?v=zyEOYl…

@kingprotty Tokio gives you `block_in_place` which you should use if you can guarantee you aren't multiplexing async code in a single task.

@kingprotty The issue (and why we don't do this) is Tokio doesn't know if there is more work to do on the current thread. Unless we tell users to *only* use Tokio primitives, anyone could use `futures::select!` and a blocking call would prevent the other branches from running.

Spur of the moment, but I had a neat idea for efficient blocking (i.e. file IO) in multi-threaded asynchronous runtimes. This is more of a note to myself and other runtime implementors. 1/n

@joshbuddy Which GPT model?

@joshbuddy ChatGPT. Totally serious. It works great (at least GPT-4 does) and it is interactive.

If you are looking for an experienced engineer who has built a lot of core OSS libraries, give Lucio a ping ->

A hardcore engineer shouldn’t let “locked doors” stop them from getting to the office.

Hmm. We sure it isn’t a “hardcore engineer” challenge?

@jasoncwarner Does that work? Huge if true.

I find myself reaching for Rust now in places I would have previously used Python. Impressively versatile and what a gift to avoid the 🐍!

@davidtolnay Wasmtime isn’t too hard to figure out. I played with it a bunch following examples.

Anyone on here familiar with Wasmtime and would be willing to collaborate on a precompiled proc macros RFC based around `--target=wasm32-macro`?