Expand description
§sack
A lock-free sack data structure and a WakerSet
for waking multiple tasks at once.
§Overview
This crate provides two data structures:
Sack<T>
: A concurrent, lock-free sack that supports adding and draining items.WakerSet
: A set of wakers that can be woken all at once.
Sack<T>
is implemented as a lock-free, singly-linked list, and WakerSet
is a wrapper around a Sack<Waker>
.
Generally this provides better performance than Mutex<Vec<T>>
for small numbers of entries. However it can vary depending on the specific use case and access patterns. Always benchmark your code to find the best solution for your particular scenario.
§Usage
Add this to your Cargo.toml
:
[dependencies]
sack = "0.1.0"
§Sack
Here is an example of how to use Sack
in a multi-producer, single-consumer scenario:
use sack::Sack;
use std::sync::Arc;
use std::thread;
let sack = Arc::new(Sack::new());
// Spawn a producer thread.
let producer = {
let sack = Arc::clone(&sack);
thread::spawn(move || {
for i in 0..10 {
sack.add(i);
}
})
};
// Wait for the producer to finish.
producer.join().unwrap();
// Drain the sack and collect the items.
let mut items: Vec<_> = sack.drain().collect();
items.sort();
assert_eq!(items, (0..10).collect::<Vec<_>>());
§WakerSet
Here is an example of how to use WakerSet
to wake up multiple tasks:
use sack::WakerSet;
use std::sync::Arc;
use std::task::{Wake, Waker};
// Create a new WakeSet.
let wake_set = Arc::new(WakerSet::new());
// Add the waker to the set.
wake_set.add_by_ref(Waker::noop());
// Wake all wakers in the set.
assert_eq!(wake_set.wake_all(), 1);
§API
The API is documented on docs.rs.
§License
This project is licensed under the MIT license. A lock-free data structure.
This crate provides a Sack<T>
type, which is a concurrent, lock-free
collection that supports adding and draining items. See Sack<T>
for more
details.
This crate also provides a WakerSet
type, which is a set of wakers that can
be woken all at once. This is useful for implementing synchronization
primitives that need to wake up multiple tasks.