Async::IO

Async::IO provides builds on async and provides asynchronous wrappers for IO, Socket, and related classes.

Installation

Add this line to your application's Gemfile:

gem 'async-io'

And then execute:

$ bundle

Or install it yourself as:

$ gem install async-io

Usage

Please browse the source code index or refer to the guides below.

Timeouts

Timeouts add a temporal limit to the execution of your code. If the IO doesn't respond in time, it will fail. Timeouts are high level concerns and you generally shouldn't use them except at the very highest level of your program.

message = task.with_timeout(5) do
  begin
    peer.read
  rescue Async::TimeoutError
    nil # The timeout was triggered.
  end
end

Any yield operation can cause a timeout to trigger. Non-async functions might not timeout because they are outside the scope of async.

Wrapper Timeouts

Asynchronous operations may block forever. You can assign a per-wrapper operation timeout duration. All asynchronous operations will be bounded by this timeout.

peer.timeout = 1
peer.read # If this takes more than 1 second, Async::TimeoutError will be raised.

The benefit of this approach is that it applies to all operations. Typically, this would be configured by the user, and set to something pretty high, e.g. 120 seconds.

Reading Characters

This example shows how to read one character at a time as the user presses it on the keyboard, and echos it back out as uppercase:

require 'async'
require 'async/io/stream'
require 'io/console'

$stdin.raw!
$stdin.echo = false

Async do |task|
  stdin = Async::IO::Stream.new(
    Async::IO::Generic.new($stdin)
  )

  while character = stdin.read(1)
    $stdout.write character.upcase
  end
end

Deferred Buffering

Async::IO::Stream.new(..., deferred:true) creates a deferred stream which increases latency slightly, but reduces the number of total packets sent. It does this by combining all calls Stream#flush within a single iteration of the reactor. This is typically more useful on the client side, but can also be useful on the server side when individual packets have high latency. It should be preferable to send one 100 byte packet than 10x 10 byte packets.

Servers typically only deal with one request per iteartion of the reactor so it's less useful. Clients which make multiple requests can benefit significantly e.g. HTTP/2 clients can merge many requests into a single packet. Because HTTP/2 recommends disabling Nagle's algorithm, this is often beneficial.

Contributing

We welcome contributions to this project.

Fork it.
Create your feature branch (git checkout -b my-new-feature).
Commit your changes (git commit -am 'Add some feature').
Push to the branch (git push origin my-new-feature).
Create new Pull Request.

Developer Certificate of Origin

This project uses the Developer Certificate of Origin. All contributors to this project must agree to this document to have their contributions accepted.

Contributor Covenant

This project is governed by the Contributor Covenant. All contributors and participants agree to abide by its terms.