Feather Runtime

Hi! Feather Runtime is the engine that powers Feather. I built it because I wanted a web server that feels as simple as writing synchronous Rust, but can still handle thousands of connections without breaking a sweat. If you’re tired of fighting with async/await or just want to see how far coroutines can take you in Rust, you’re in the right place.

It replaces tiny-http with a modern, coroutine-based runtime—no async/await required.

🚀 Features

• Coroutines, Not Threads: Thanks to the may crate, every connection gets its own coroutine (a green thread). This means you can handle a ton of traffic without your server falling over.
• Non-blocking I/O: All sockets are non-blocking, so the server stays snappy even when things get busy.
• No async/await Headaches: Just write normal Rust code. Feather-Runtime takes care of the scheduling magic behind the scenes.
• Graceful Shutdown: Hit Ctrl+C and the server shuts down cleanly, no weird hangs.
• Dynamic HTTP Responses: Build and send responses however you want—no fuss.
• Buffered I/O: Uses BufReader and BufWriter for speed.
• Custom Socket Tuning: Want to tweak backlog, nodelay, or buffer sizes? It’s all there via socket2.
• Extensible: It’s the engine for Feather, but you can use it for your own experiments too.

🛠️ How It Works

I wanted something that “just works” for high concurrency, but doesn’t make you write async spaghetti. Here’s the gist:

• Coroutines, Not Threads: Instead of a thread per connection, every connection gets a coroutine (thanks, may). Coroutines are super lightweight, so you can have thousands running at once.
• Non-blocking, Event-driven: Sockets are non-blocking. When a request comes in, it’s handed to a coroutine. If that coroutine needs to wait for I/O, it just yields and lets others do their thing. No wasted CPU, no blocking the whole server.
• Message Queues: Requests go into a queue, and coroutines pick them up, process, and respond. This keeps things smooth and scalable.
• No async/await, No Lifetimes: Write normal Rust. No async, no lifetimes, no pinning. The runtime handles all the tricky stuff.
• Socket Tuning: Want to tweak how the server listens? Use socket2 to set backlog, nodelay, buffer sizes, etc.

In summary:

Every request gets its own coroutine. You can spawn background tasks or run blocking code, and Feather-Runtime will keep things fast and responsive—no async/await or lifetime headaches.

📦 Example Usage

Here’s a minimal example of using Feather-Runtime directly (normally, you use it via Feather):

use feather_runtime::runtime::engine::Engine;
use feather_runtime::http::{Request, Response};

fn main() {
    let engine = Engine::new("127.0.0.1:5050");
    engine.start();
    engine.for_each(|req: &mut Request| {
        let mut res = Response::default();
        res.send_text("Hello from Feather-Runtime!");
        res
    }).unwrap();
}

🤝 For Contributors

If you're contributing to Feather but don't want to mess with low-level server internals, you can mostly ignore this subcrate. Feather-Runtime is designed to handle the core HTTP processing while Feather itself provides higher-level abstractions. If you have a feature request or a problem, open an issue.

📚 Learn More

• Feather Main Repo
• may Crate
• socket2 Crate

License

Feather-Runtime is MIT licensed. See LICENSE.