#worker #worker-thread #multi-threading #run-time

employees

A small runtime that hides all the boilerplate when using threads

7 releases

0.1.0 Sep 23, 2024
0.0.6 Apr 8, 2024
0.0.4 Dec 15, 2023
0.0.3 Nov 21, 2023

#247 in Concurrency

Download history 2/week @ 2024-06-28 16/week @ 2024-07-05 166/week @ 2024-09-20 27/week @ 2024-09-27 6/week @ 2024-10-04 4/week @ 2024-10-11

203 downloads per month

MIT license

67KB
874 lines

employees

A small and lightweight crate to hide most of the burden of setting up long-living threads.

Philosophy

This crate sees threads as unique entities called workers which may (or may not) live as long as the program lives. Workers are spawned in Runtimes that manage them.

Usage

Here's a small example that spawns a worker that prints "Hello, World!" every 100ms for 1 second.

struct WorkerThatPrints;
impl Worker for WorkerThatPrints {
    fn on_update(&mut self) -> ControlFlow {
        println!("Hello, World!");
        std::thread::sleep(Duration::from_millis(100));
        ControlFlow::Continue
    }
}

let mut runtime = Runtime::new();

runtime.launch(WorkerThatPrints);
std::thread::sleep(Duration::from_secs(1));

See the full documentation for more in depth details.

Should I use employees?

Concurrency

employees is built with CPU-bound concurrency in mind.

"But wait, arent't async runtimes built for that purpose?" you might ask, and you'll be 100% right.

However, async has drawbacks which employees hasn't:

That said, there are usecases where async runtime will be better such as I/O bound tasks (web servers, ect.) or concurrent tasks on single threaded platforms.

Paralellism

Again, employees is built with concurrency in mind. While it is possible to build a work stealing thread pool from employees to parallelize some kind of computation, other crates such as rayon add a ton of utilities speficically made for that purpose which is why, for most paralellism usecases, rayon-like crates will be a better choice.

License

Licensed under the terms of MIT license. See LICENSE for details.

Dependencies

~0.2–6.5MB
~37K SLoC