Struct Endpoint

pub struct Endpoint<S, T> { /* private fields */ }

A Future for running both a client and a server at the same time.

The client part will be provided to the ServiceWithClient::handle_request and ServiceWithClient::handle_notification methods, so that the server can send back requests and notifications as part of its handling duties. You may also access the client with the client() method if you want to send additional requests.

The returned future needs to be spawned onto a task in order to actually run the server (and the client). It will run until the stream is closed; if the stream encounters an error, the future will propagate it and terminate.

use std::io;
use rmp_rpc::ServiceWithClient;
use std::net::SocketAddr;
use tokio::net::TcpListener;
use tokio_util::compat::TokioAsyncReadCompatExt;

struct MyService;
impl ServiceWithClient for MyService {
// ...
}

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr: SocketAddr = "127.0.0.1:54321".parse().unwrap();

    // Here's the simplest version: we listen for incoming TCP connections and run an
    // endpoint on each one.
    let server = async {
        let mut listener = TcpListener::bind(&addr).await?;
        loop {
            // Each time the listener finds a new connection, start up an endpoint to handle
            // it.
            let (socket, _) = listener.accept().await?;
            if let Err(e) = Endpoint::new(socket.compat(), MyService).await {
                println!("error on endpoint {}", e);
            }
        }
    };

    // Uncomment this to run the server on the tokio event loop. This is blocking.
    // Press ^C to stop
    // tokio::run(server);

    // Here's an alternative, where we take a handle to the client and spawn the endpoint
    // on its own task.
    let addr: SocketAddr = "127.0.0.1:65432".parse().unwrap();
    let server = async {
        let mut listener = TcpListener::bind(&addr).await?;
        loop {
            let (socket, _) = listener.accept().await?;
            let end = Endpoint::new(socket.compat(), MyService);
            let client = end.client();

            // Spawn the endpoint. It will do its own thing, while we can use the client
            // to send requests.
            tokio::spawn(end);

            // Send a request with method name "hello" and argument "world!".
            match client.request("hello", &["world!".into()]).await {
                Ok(response) => println!("{:?}", response),
                Err(e) => println!("got an error: {:?}", e),
            };
            // We're returning the future that came from `client.request`. This means that
            // `server` (and therefore our entire program) will terminate once the
            // response is received and the messages are printed. If you wanted to keep
            // the endpoint running even after the response is received, you could
            // (instead of spawning `end` on its own task) `join` the two futures (i.e.
            // `end` and the one returned by `client.request`).
        }
    };

    // Uncomment this to run the server on the tokio event loop. This is blocking.
    // Press ^C to stop
    // tokio::run(server);

    Ok(())
}

Implementations

impl<S: ServiceWithClient + Unpin, T: AsyncRead + AsyncWrite> Endpoint<S, T>

pub fn new(stream: T, service: S) -> Self

Creates a new Endpoint on stream, using service to handle requests and notifications.

pub fn client(&self) -> Client

Returns a handle to the client half of this Endpoint, which can be used for sending requests and notifications.

Trait Implementations

impl<S: ServiceWithClient + Unpin, T: AsyncRead + AsyncWrite> Future for Endpoint<S, T>

type Output = Result<(), Error>

The type of value produced on completion.

fn poll(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Self::Output>

Attempts to resolve the future to a final value, registering the current task for wakeup if the value is not yet available.