mco is a high-performance library for programming stackful coroutines with which you can easily develop and maintain massive concurrent programs. It can be thought as the Rust version of the popular [Goroutine][go].
- Most commonly used APIs
- The current version support stack expansion
- Elegant coding,No need for async await
- Simple concurrency(CSP model), learning Golang
- Default MAX 6MB Stack Size
- many std like API
Initial code frok from May and we add Many improvements(Inspired by Golang, parking_lot and crossbeam) and more...
mco Powerful standard library
mco/std/queue
Basic queue data structuresmco/std/sync
IncludesMutex/RwLock/WaitGroup/Semphore/chan!()/chan!(1000)
...and more..mco/std/defer
Defers evaluation of a block of code until the end of the scope.mco/std/map
Provides the same concurrency map as Golang, withSyncHashMap
andSyncBtreeMap
.It is suitable for concurrent environments with too many reads and too few writesmco/std/vec
Provides the same concurrency vecmco/std/time
Improve the implementation of a high performance timemco/std/lazy
Thread/coroutine safe global variable,Lazy struct,OnceCell
Crates based on mco implementation
- mco-http High-performance coroutine HTTP server and client
- cdbc Database Drivers include mysql, Postgres, AND SQLite
- fast_log High-performance log impl
- mco-redis Redis client for mco
- mco-redis-rs fork from
redis-rs
Replace only TcpStream with MCO ::TcpStream - mco-rpc rpc server/client. support bincode/json rpc
-
The stackful coroutine implementation is based on [generator][generator];
-
Support schedule on a configurable number of threads for multi-core systems;
-
Support coroutine version of a local storage ([CLS][cls]);
-
Support efficient asynchronous network I/O;
-
Support efficient timer management;
-
Support standard synchronization primitives, a semaphore, an MPMC channel, etc;
-
Support cancellation of coroutines;
-
Support graceful panic handling that will not affect other coroutines;
-
Support scoped coroutine creation;
-
Support general selection for all the coroutine API;
-
All the coroutine API are compatible with the standard library semantics;
-
All the coroutine API can be safely called in multi-threaded context;
-
Both stable, beta, and nightly channels are supported;
-
x86_64 GNU/Linux, x86_64 Windows, x86_64 Mac, aarch64 Linux OS are supported.
-
Support High performance chan(like golang)
-
Support WaitGroup Support(like golang)
-
Support defer!() (like golang)
-
Support Rustls
-
Support Time (like golang)
-
Support error/err!() (like golang)
-
Support select match Ok(v)/Err(e) (like golang)
-
Support Lazy/OnceCell
-
Support SyncMap(like golang)
-
Support Ticker(like golang)
mco = "0.1"
A naive echo server implemented with mco:
#[macro_use]
extern crate mco;
use mco::net::TcpListener;
use std::io::{Read, Write};
fn main() {
let listener = TcpListener::bind("127.0.0.1:8000").unwrap();
while let Ok((mut stream, _)) = listener.accept() {
go!(move || {
let mut buf = vec![0; 1024 * 16]; // alloc in heap!
while let Ok(n) = stream.read(&mut buf) {
if n == 0 {
break;
}
stream.write_all(&buf[0..n]).unwrap();
}
});
}
}
There is a detailed [document][caveat] that describes mco's main restrictions. In general, there are four things you should follow when writing programs that use coroutines:
- Don't call thread-blocking API (It will hurt the performance);
- Carefully use Thread Local Storage (access TLS in coroutine might trigger undefined behavior).
It's considered unsafe with the following pattern:
set_tls(); // Or another coroutine API that would cause scheduling: yield_now(); use_tls();but it's safe if your code is not sensitive about the previous state of TLS. Or there is no coroutines scheduling between set TLS and use TLS.
- Don't run CPU bound tasks for long time, but it's ok if you don't care about fairness;
- Don't run thread::sleep() on mco coroutine
- In most modern operating systems, when starting a process, the standard Thread stack size is usually 8 MB, and mco provides a maximum stack space of 6MB. Typically, operating systems load memory pages on demand, such as starting about 1 million processes on my Mac/Unix system, which requires 15GB of memory space.
- Don't exceed the coroutine stack. There is a guard page for each coroutine stack. When stack overflow occurs, it will trigger segment fault error.
Note:
The first three rules are common when using cooperative asynchronous libraries in Rust. Even using a futures-based system also have these limitations. So what you should really focus on is a coroutine stack size, make sure it's big enough for your applications.
mco::config().set_stack_size(6*1024*1024);
- We are in urgent need of financial support and cooperation, welcome to contact us
- email: [email protected]
- wechat: zxj347284221
捐赠
联系方式(添加好友请备注'mco') 微信群:先加微信,然后拉进群