timely_communication 0.14.0

Communication layer for timely dataflow
Documentation

A simple communication infrastructure providing typed exchange channels.

This crate is part of the timely dataflow system, used primarily for its inter-worker communication. It may be independently useful, but it is separated out mostly to make clear boundaries in the project.

Threads are spawned with an allocator::Generic, whose allocate method returns a pair of several send endpoints and one receive endpoint. Messages sent into a send endpoint will eventually be received by the corresponding worker, if it receives often enough. The point-to-point channels are each FIFO, but with no fairness guarantees.

To be communicated, a type must implement the [Bytesable] trait.

Channel endpoints also implement a lower-level push and pull interface (through the [Push] and [Pull] traits), which is used for more precise control of resources.

Examples

use timely_communication::{Allocate, Bytesable};

/// A wrapper that indicates `bincode` as the serialization/deserialization strategy.
pub struct Message {
/// Text contents.
pub payload: String,
}

impl Bytesable for Message {
fn from_bytes(bytes: timely_bytes::arc::Bytes) -> Self {
Message { payload: std::str::from_utf8(&bytes[..]).unwrap().to_string() }
}

fn length_in_bytes(&self) -> usize {
self.payload.len()
}

fn into_bytes<W: ::std::io::Write>(&self, writer: &mut W) {
writer.write_all(self.payload.as_bytes()).unwrap();
}
}

fn main() {

// extract the configuration from user-supplied arguments, initialize the computation.
let config = timely_communication::Config::from_args(std::env::args()).unwrap();
let guards = timely_communication::initialize(config, |mut allocator| {

println!("worker {} of {} started", allocator.index(), allocator.peers());

// allocates a pair of senders list and one receiver.
let (mut senders, mut receiver) = allocator.allocate(0);

// send typed data along each channel
for i in 0 .. allocator.peers() {
senders[i].send(Message { payload: format!("hello, {}", i)});
senders[i].done();
}

// no support for termination notification,
// we have to count down ourselves.
let mut received = 0;
while received < allocator.peers() {

allocator.receive();

if let Some(message) = receiver.recv() {
println!("worker {}: received: <{}>", allocator.index(), message.payload);
received += 1;
}

allocator.release();
}

allocator.index()
});

// computation runs until guards are joined or dropped.
if let Ok(guards) = guards {
for guard in guards.join() {
println!("result: {:?}", guard);
}
}
else { println!("error in computation"); }
}

This should produce output like:

worker 0 started
worker 1 started
worker 0: received: <hello, 0>
worker 1: received: <hello, 1>
worker 0: received: <hello, 0>
worker 1: received: <hello, 1>
result: Ok(0)
result: Ok(1)