Expand description

Zero-cost Futures in Rust

This library is an implementation of futures in Rust which aims to provide a robust implementation of handling asynchronous computations, ergonomic composition and usage, and zero-cost abstractions over what would otherwise be written by hand.

Futures are a concept for an object which is a proxy for another value that may not be ready yet. For example issuing an HTTP request may return a future for the HTTP response, as it probably hasn’t arrived yet. With an object representing a value that will eventually be available, futures allow for powerful composition of tasks through basic combinators that can perform operations like chaining computations, changing the types of futures, or waiting for two futures to complete at the same time.

You can find extensive tutorials and documentations at https://tokio.rs for both this crate (asynchronous programming in general) as well as the Tokio stack to perform async I/O with.

Installation

Add this to your Cargo.toml:

[dependencies]
futures = "0.1"

Examples

Let’s take a look at a few examples of how futures might be used:

extern crate futures;

use std::io;
use std::time::Duration;
use futures::prelude::*;
use futures::future::Map;

// A future is actually a trait implementation, so we can generically take a
// future of any integer and return back a future that will resolve to that
// value plus 10 more.
//
// Note here that like iterators, we're returning the `Map` combinator in
// the futures crate, not a boxed abstraction. This is a zero-cost
// construction of a future.
fn add_ten<F>(future: F) -> Map<F, fn(i32) -> i32>
    where F: Future<Item=i32>,
{
    fn add(a: i32) -> i32 { a + 10 }
    future.map(add)
}

// Not only can we modify one future, but we can even compose them together!
// Here we have a function which takes two futures as input, and returns a
// future that will calculate the sum of their two values.
//
// Above we saw a direct return value of the `Map` combinator, but
// performance isn't always critical and sometimes it's more ergonomic to
// return a trait object like we do here. Note though that there's only one
// allocation here, not any for the intermediate futures.
fn add<'a, A, B>(a: A, b: B) -> Box<Future<Item=i32, Error=A::Error> + 'a>
    where A: Future<Item=i32> + 'a,
          B: Future<Item=i32, Error=A::Error> + 'a,
{
    Box::new(a.join(b).map(|(a, b)| a + b))
}

// Futures also allow chaining computations together, starting another after
// the previous finishes. Here we wait for the first computation to finish,
// and then decide what to do depending on the result.
fn download_timeout(url: &str,
                    timeout_dur: Duration)
                    -> Box<Future<Item=Vec<u8>, Error=io::Error>> {
    use std::io;
    use std::net::{SocketAddr, TcpStream};

    type IoFuture<T> = Box<Future<Item=T, Error=io::Error>>;

    // First thing to do is we need to resolve our URL to an address. This
    // will likely perform a DNS lookup which may take some time.
    let addr = resolve(url);

    // After we acquire the address, we next want to open up a TCP
    // connection.
    let tcp = addr.and_then(|addr| connect(&addr));

    // After the TCP connection is established and ready to go, we're off to
    // the races!
    let data = tcp.and_then(|conn| download(conn));

    // That all might take awhile, though, so let's not wait too long for it
    // to all come back. The `select` combinator here returns a future which
    // resolves to the first value that's ready plus the next future.
    //
    // Note we can also use the `then` combinator which is similar to
    // `and_then` above except that it receives the result of the
    // computation, not just the successful value.
    //
    // Again note that all the above calls to `and_then` and the below calls
    // to `map` and such require no allocations. We only ever allocate once
    // we hit the `Box::new()` call at the end here, which means we've built
    // up a relatively involved computation with only one box, and even that
    // was optional!

    let data = data.map(Ok);
    let timeout = timeout(timeout_dur).map(Err);

    let ret = data.select(timeout).then(|result| {
        match result {
            // One future succeeded, and it was the one which was
            // downloading data from the connection.
            Ok((Ok(data), _other_future)) => Ok(data),

            // The timeout fired, and otherwise no error was found, so
            // we translate this to an error.
            Ok((Err(_timeout), _other_future)) => {
                Err(io::Error::new(io::ErrorKind::Other, "timeout"))
            }

            // A normal I/O error happened, so we pass that on through.
            Err((e, _other_future)) => Err(e),
        }
    });
    return Box::new(ret);

    fn resolve(url: &str) -> IoFuture<SocketAddr> {
        // ...
    }

    fn connect(hostname: &SocketAddr) -> IoFuture<TcpStream> {
        // ...
    }

    fn download(stream: TcpStream) -> IoFuture<Vec<u8>> {
        // ...
    }

    fn timeout(stream: Duration) -> IoFuture<()> {
        // ...
    }
}

Some more information can also be found in the README for now, but otherwise feel free to jump in to the docs below!

Re-exports

pub use future::Future;
pub use future::IntoFuture;
pub use stream::Stream;
pub use sink::Sink;

Modules

Executors

Futures

A “prelude” for crates using the futures crate.

Asynchronous sinks

Asynchronous streams

Future-aware synchronization

Tasks used to drive a future computation

Future-aware single-threaded synchronization

Macros

A macro to create a static of type LocalKey

A macro for extracting the successful type of a Poll<T, E>.

Enums

Return type of future, indicating whether a value is ready or not.

The result of an asynchronous attempt to send a value to a sink.

Type Definitions

Return type of the Future::poll method, indicates whether a future’s value is ready or not.

Return type of the Sink::start_send method, indicating the outcome of a send attempt. See AsyncSink for more details.