Module: EventMachine

Defined in:
lib/em/future.rb,
lib/em/pool.rb,
lib/em/queue.rb,
lib/em/timers.rb,
lib/em/channel.rb,
lib/em/version.rb,
lib/em/callback.rb,
lib/em/iterator.rb,
lib/em/resolver.rb,
lib/em/streamer.rb,
lib/em/processes.rb,
lib/em/protocols.rb,
lib/em/spawnable.rb,
lib/em/tick_loop.rb,
lib/eventmachine.rb,
lib/em/completion.rb,
lib/em/connection.rb,
lib/em/deferrable.rb,
lib/em/file_watch.rb,
lib/em/process_watch.rb,
lib/em/protocols/stomp.rb,
lib/em/protocols/socks4.rb,
lib/em/protocols/tcptest.rb,
lib/em/threaded_resource.rb,
lib/em/protocols/memcache.rb,
lib/em/protocols/saslauth.rb,
lib/em/protocols/linetext2.rb,
lib/em/protocols/postgres3.rb,
lib/em/protocols/httpclient.rb,
lib/em/protocols/smtpclient.rb,
lib/em/protocols/smtpserver.rb,
lib/em/protocols/httpclient2.rb,
lib/em/protocols/line_and_text.rb,
lib/em/protocols/line_protocol.rb,
lib/em/protocols/object_protocol.rb,
lib/em/protocols/header_and_content.rb

Overview

--

Author:: Francis Cianfrocca (gmail: blackhedd) Homepage:: http://rubyeventmachine.com Date:: 15 Nov 2006

See EventMachine and EventMachine::Connection for documentation and usage examples.


Copyright (C) 2006-07 by Francis Cianfrocca. All Rights Reserved. Gmail: blackhedd

This program is free software; you can redistribute it and/or modify it under the terms of either: 1) the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version; or 2) Ruby's License.

See the file COPYING for complete licensing information.


Defined Under Namespace

Modules: DNS, Deferrable, Protocols Classes: Channel, Completion, Connection, DefaultDeferrable, DeferrableChildProcess, FileNotFoundException, FileStreamer, FileWatch, Iterator, PeriodicTimer, Pool, ProcessWatch, Queue, SpawnedProcess, ThreadedResource, TickLoop, Timer

Constant Summary collapse

VERSION =
"1.0.0"
P =

Alias for Protocols

EventMachine::Protocols

Class Attribute Summary collapse

Class Method Summary collapse

Class Attribute Details

.reactor_threadThread (readonly)

Exposed to allow joining on the thread, when run in a multithreaded environment. Performing other actions on the thread has undefined semantics (read: a dangerous endevor).

Returns:

  • (Thread)


79
80
81
# File 'lib/eventmachine.rb', line 79

def reactor_thread
  @reactor_thread
end

.threadpool_sizeNumber

Size of the EventMachine.defer threadpool (defaults to 20)

Returns:

  • (Number)


1067
1068
1069
# File 'lib/eventmachine.rb', line 1067

def threadpool_size
  @threadpool_size
end

Class Method Details

.add_periodic_timer(*args, &block) ⇒ Object

Adds a periodic timer to the event loop. It takes the same parameters as the one-shot timer method, add_timer. This method schedules execution of the given block repeatedly, at intervals of time at least as great as the number of seconds given in the first parameter to the call.

Examples:

Write a dollar-sign to stderr every five seconds, without blocking


EventMachine.run {
  EventMachine.add_periodic_timer( 5 ) { $stderr.write "$" }
}

Parameters:

  • delay (Integer)

    Delay in seconds

See Also:



348
349
350
351
352
353
# File 'lib/eventmachine.rb', line 348

def self.add_periodic_timer *args, &block
  interval = args.shift
  code = args.shift || block

  EventMachine::PeriodicTimer.new(interval, code)
end

.add_shutdown_hook(&block) ⇒ Object

Adds a block to call as the reactor is shutting down.

These callbacks are called in the reverse order to which they are added.

Examples:

Scheduling operations to be run when EventMachine event loop is stopped


EventMachine.run do
  EventMachine.add_shutdown_hook { puts "b" }
  EventMachine.add_shutdown_hook { puts "a" }
  EventMachine.stop
end

# Outputs:
#   a
#   b


288
289
290
# File 'lib/eventmachine.rb', line 288

def self.add_shutdown_hook &block
  @tails << block
end

.add_timer(*args, &block) ⇒ Object

Adds a one-shot timer to the event loop. Call it with one or two parameters. The first parameters is a delay-time expressed in seconds (not milliseconds). The second parameter, if present, must be an object that responds to :call. If 2nd parameter is not given, then you can also simply pass a block to the method call.

This method may be called from the block passed to run or from any callback method. It schedules execution of the proc or block passed to it, after the passage of an interval of time equal to at least the number of seconds specified in the first parameter to the call.

add_timer is a non-blocking method. Callbacks can and will be called during the interval of time that the timer is in effect. There is no built-in limit to the number of timers that can be outstanding at any given time.

Examples:

Setting a one-shot timer with EventMachine


EventMachine.run {
  puts "Starting the run now: #{Time.now}"
  EventMachine.add_timer 5, proc { puts "Executing timer event: #{Time.now}" }
  EventMachine.add_timer(10) { puts "Executing timer event: #{Time.now}" }
}

Parameters:

  • delay (Integer)

    Delay in seconds

See Also:



320
321
322
323
324
325
326
327
328
329
# File 'lib/eventmachine.rb', line 320

def self.add_timer *args, &block
  interval = args.shift
  code = args.shift || block
  if code
    # check too many timers!
    s = add_oneshot_timer((interval.to_f * 1000).to_i)
    @timers[s] = code
    s
  end
end

.attach(io, handler = nil, *args, &blk) ⇒ Object

Attaches an IO object or file descriptor to the eventloop as a regular connection. The file descriptor will be set as non-blocking, and EventMachine will process receive_data and send_data events on it as it would for any other connection.

To watch a fd instead, use watch, which will not alter the state of the socket and fire notify_readable and notify_writable events instead.



729
730
731
# File 'lib/eventmachine.rb', line 729

def EventMachine::attach io, handler=nil, *args, &blk
  attach_io io, false, handler, *args, &blk
end

.bind_connect(bind_addr, bind_port, server, port = nil, handler = nil, *args) ⇒ Object

This method is like connect, but allows for a local address/port to bind the connection to.

See Also:



649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
# File 'lib/eventmachine.rb', line 649

def self.bind_connect bind_addr, bind_port, server, port=nil, handler=nil, *args
  begin
    port = Integer(port)
  rescue ArgumentError, TypeError
    # there was no port, so server must be a unix domain socket
    # the port argument is actually the handler, and the handler is one of the args
    args.unshift handler if handler
    handler = port
    port = nil
  end if port

  klass = klass_from_handler(Connection, handler, *args)

  s = if port
        if bind_addr
          bind_connect_server bind_addr, bind_port.to_i, server, port
        else
          connect_server server, port
        end
      else
        connect_unix_server server
      end

  c = klass.new s, *args
  @conns[s] = c
  block_given? and yield c
  c
end

.Callback(object, method) ⇒ <#call> .Callback(object) ⇒ <#call> .Callback(&block) ⇒ <#call>

Utility method for coercing arguments to an object that responds to :call. Accepts an object and a method name to send to, or a block, or an object that responds to :call.

Examples:

EventMachine.Callback used with a block. Returns that block.


cb = EventMachine.Callback do |msg|
  puts(msg)
end
# returned object is a callable
cb.call('hello world')

EventMachine.Callback used with an object (to be more specific, class object) and a method name, returns an object that responds to #call


cb = EventMachine.Callback(Object, :puts)
# returned object is a callable that delegates to Kernel#puts (in this case Object.puts)
cb.call('hello world')

EventMachine.Callback used with an object that responds to #call. Returns the argument.


cb = EventMachine.Callback(proc{ |msg| puts(msg) })
# returned object is a callable
cb.call('hello world')

Overloads:

  • .Callback(object, method) ⇒ <#call>

    Wraps method invocation on object into an object that responds to #call that proxies all the arguments to that method

    Parameters:

    • Object (Object)

      to invoke method on

    • Method (Symbol)

      name

    Returns:

    • (<#call>)

      An object that responds to #call that takes any number of arguments and invokes method on object with those arguments

  • .Callback(object) ⇒ <#call>

    Returns callable object as is, without any coercion

    Parameters:

    • An (<#call>)

      object that responds to #call

    Returns:

    • (<#call>)

      Its argument

  • .Callback(&block) ⇒ <#call>

    Returns block passed to it without any coercion

    Returns:

    • (<#call>)

      Block passed to this method

Returns:

  • (<#call>)

Raises:

  • (ArgumentError)

    When argument doesn't respond to #call, method name is missing or when invoked without arguments and block isn't given



47
48
49
50
51
52
53
54
55
56
57
# File 'lib/em/callback.rb', line 47

def self.Callback(object = nil, method = nil, &blk)
  if object && method
    lambda { |*args| object.__send__ method, *args }
  else
    if object.respond_to? :call
      object
    else
      blk || raise(ArgumentError)
    end # if
  end # if
end

.cancel_timer(timer_or_sig) ⇒ Object

Cancel a timer (can be a callback or an Timer instance).

Parameters:

  • timer_or_sig (#cancel, #call)

    A timer to cancel

See Also:



360
361
362
363
364
365
366
# File 'lib/eventmachine.rb', line 360

def self.cancel_timer timer_or_sig
  if timer_or_sig.respond_to? :cancel
    timer_or_sig.cancel
  else
    @timers[timer_or_sig] = false if @timers.has_key?(timer_or_sig)
  end
end

.connect(server, port = nil, handler = nil, *args, &blk) ⇒ Object

Initiates a TCP connection to a remote server and sets up event handling for the connection. connect requires event loop to be running (see run).

connect takes the IP address (or hostname) and port of the remote server you want to connect to. It also takes an optional handler (a module or a subclass of Connection) which you must define, that contains the callbacks that will be invoked by the event loop on behalf of the connection.

Learn more about connection lifecycle callbacks in the EventMachine tutorial and Connection lifecycle guide.

Examples:


# Here's a program which connects to a web server, sends a naive
# request, parses the HTTP header of the response, and then
# (antisocially) ends the event loop, which automatically drops the connection
# (and incidentally calls the connection's unbind method).
module DumbHttpClient
  def post_init
    send_data "GET / HTTP/1.1\r\nHost: _\r\n\r\n"
    @data = ""
    @parsed = false
  end

  def receive_data data
    @data << data
    if !@parsed and @data =~ /[\n][\r]*[\n]/m
      @parsed = true
      puts "RECEIVED HTTP HEADER:"
      $`.each {|line| puts ">>> #{line}" }

      puts "Now we'll terminate the loop, which will also close the connection"
      EventMachine::stop_event_loop
    end
  end

  def unbind
    puts "A connection has terminated"
  end
end

EventMachine.run {
  EventMachine.connect "www.bayshorenetworks.com", 80, DumbHttpClient
}
puts "The event loop has ended"

Defining protocol handler as a class


class MyProtocolHandler < EventMachine::Connection
  def initialize *args
    super
    # whatever else you want to do here
  end

  # ...
end

Parameters:

  • server (String)

    Host to connect to

  • port (Integer) (defaults to: nil)

    Port to connect to

  • handler (Module, Class) (defaults to: nil)

    A module or class that implements connection lifecycle callbacks

See Also:



619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
# File 'lib/eventmachine.rb', line 619

def self.connect server, port=nil, handler=nil, *args, &blk
  # EventMachine::connect initiates a TCP connection to a remote
  # server and sets up event-handling for the connection.
  # It internally creates an object that should not be handled
  # by the caller. HOWEVER, it's often convenient to get the
  # object to set up interfacing to other objects in the system.
  # We return the newly-created anonymous-class object to the caller.
  # It's expected that a considerable amount of code will depend
  # on this behavior, so don't change it.
  #
  # Ok, added support for a user-defined block, 13Apr06.
  # This leads us to an interesting choice because of the
  # presence of the post_init call, which happens in the
  # initialize method of the new object. We call the user's
  # block and pass the new object to it. This is a great
  # way to do protocol-specific initiation. It happens
  # AFTER post_init has been called on the object, which I
  # certainly hope is the right choice.
  # Don't change this lightly, because accepted connections
  # are different from connected ones and we don't want
  # to have them behave differently with respect to post_init
  # if at all possible.

  bind_connect nil, nil, server, port, handler, *args, &blk
end

.connect_unix_domain(socketname, *args, &blk) ⇒ Object

Note:

UNIX sockets, as the name suggests, are not available on Microsoft Windows.

Make a connection to a Unix-domain socket. This method is simply an alias for connect, which can connect to both TCP and Unix-domain sockets. Make sure that your process has sufficient permissions to open the socket it is given.

Parameters:

  • socketname (String)

    Unix domain socket (local fully-qualified path) you want to connect to.



796
797
798
# File 'lib/eventmachine.rb', line 796

def self.connect_unix_domain socketname, *args, &blk
  connect socketname, *args, &blk
end

.connection_countInteger

Returns the total number of connections (file descriptors) currently held by the reactor. Note that a tick must pass after the 'initiation' of a connection for this number to increment. It's usually accurate, but don't rely on the exact precision of this number unless you really know EM internals.

Examples:


EventMachine.run {
  EventMachine.connect("rubyeventmachine.com", 80)
  # count will be 0 in this case, because connection is not
  # established yet
  count = EventMachine.connection_count
}

EventMachine.run {
  EventMachine.connect("rubyeventmachine.com", 80)

  EventMachine.next_tick {
    # In this example, count will be 1 since the connection has been established in
    # the next loop of the reactor.
    count = EventMachine.connection_count
  }
}

Returns:

  • (Integer)

    Number of connections currently held by the reactor.



938
939
940
# File 'lib/eventmachine.rb', line 938

def self.connection_count
  self.get_connection_count
end

.defer(op = nil, callback = nil, &blk) ⇒ Object

EventMachine.defer is used for integrating blocking operations into EventMachine's control flow. The action of defer is to take the block specified in the first parameter (the "operation") and schedule it for asynchronous execution on an internal thread pool maintained by EventMachine. When the operation completes, it will pass the result computed by the block (if any) back to the EventMachine reactor. Then, EventMachine calls the block specified in the second parameter to defer (the "callback"), as part of its normal event handling loop. The result computed by the operation block is passed as a parameter to the callback. You may omit the callback parameter if you don't need to execute any code after the operation completes.

Caveats

Note carefully that the code in your deferred operation will be executed on a separate thread from the main EventMachine processing and all other Ruby threads that may exist in your program. Also, multiple deferred operations may be running at once! Therefore, you are responsible for ensuring that your operation code is threadsafe.

Don't write a deferred operation that will block forever. If so, the current implementation will not detect the problem, and the thread will never be returned to the pool. EventMachine limits the number of threads in its pool, so if you do this enough times, your subsequent deferred operations won't get a chance to run.

Examples:


operation = proc {
  # perform a long-running operation here, such as a database query.
  "result" # as usual, the last expression evaluated in the block will be the return value.
}
callback = proc {|result|
  # do something with result here, such as send it back to a network client.
}

EventMachine.defer(operation, callback)

Parameters:

  • op (#call) (defaults to: nil)

    An operation you want to offload to EventMachine thread pool

  • callback (#call) (defaults to: nil)

    A callback that will be run on the event loop thread after operation finishes.

See Also:



1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
# File 'lib/eventmachine.rb', line 1011

def self.defer op = nil, callback = nil, &blk
  # OBSERVE that #next_tick hacks into this mechanism, so don't make any changes here
  # without syncing there.
  #
  # Running with $VERBOSE set to true gives a warning unless all ivars are defined when
  # they appear in rvalues. But we DON'T ever want to initialize @threadqueue unless we
  # need it, because the Ruby threads are so heavyweight. We end up with this bizarre
  # way of initializing @threadqueue because EventMachine is a Module, not a Class, and
  # has no constructor.

  unless @threadpool
    @threadpool = []
    @threadqueue = ::Queue.new
    @resultqueue = ::Queue.new
    spawn_threadpool
  end

  @threadqueue << [op||blk,callback]
end

.defers_finished?Boolean

Returns +true+ if all deferred actions are done executing and their callbacks have been fired.

Returns:

  • (Boolean)


1053
1054
1055
1056
1057
1058
1059
# File 'lib/eventmachine.rb', line 1053

def self.defers_finished?
  return false if @threadpool and !@all_threads_spawned
  return false if @threadqueue and not @threadqueue.empty?
  return false if @resultqueue and not @resultqueue.empty?
  return false if @threadpool and @threadqueue.num_waiting != @threadpool.size
  return true
end

.disable_proxy(from) ⇒ Object

Takes just one argument, a Connection that has proxying enabled via enable_proxy. Calling this method will remove that functionality and your connection will begin receiving data via EventMachine::Connection#receive_data again.

Parameters:

See Also:



1393
1394
1395
# File 'lib/eventmachine.rb', line 1393

def self.disable_proxy(from)
  EM::stop_proxy(from.signature)
end

.enable_proxy(from, to, bufsize = 0, length = 0) ⇒ Object

This method allows for direct writing of incoming data back out to another descriptor, at the C++ level in the reactor. This is very efficient and especially useful for proxies where high performance is required. Propogating data from a server response all the way up to Ruby, and then back down to the reactor to be sent back to the client, is often unnecessary and incurs a significant performance decrease.

The two arguments are instance of Connection subclasses, 'from' and 'to'. 'from' is the connection whose inbound data you want relayed back out. 'to' is the connection to write it to.

Once you call this method, the 'from' connection will no longer get receive_data callbacks from the reactor, except in the case that 'to' connection has already closed when attempting to write to it. You can see in the example, that proxy_target_unbound will be called when this occurs. After that, further incoming data will be passed into receive_data as normal.

Note also that this feature supports different types of descriptors: TCP, UDP, and pipes. You can relay data from one kind to another, for example, feed a pipe from a UDP stream.

Examples:


module ProxyConnection
  def initialize(client, request)
    @client, @request = client, request
  end

  def post_init
    EM::enable_proxy(self, @client)
  end

  def connection_completed
    send_data @request
  end

  def proxy_target_unbound
    close_connection
  end

  def unbind
    @client.close_connection_after_writing
  end
end

module ProxyServer
  def receive_data(data)
    (@buf ||= "") << data
    if @buf =~ /\r\n\r\n/ # all http headers received
      EventMachine.connect("10.0.0.15", 80, ProxyConnection, self, data)
    end
  end
end

EventMachine.run {
  EventMachine.start_server("127.0.0.1", 8080, ProxyServer)
}

Parameters:

  • from (EventMachine::Connection)

    Source of data to be proxies/streamed.

  • to (EventMachine::Connection)

    Destination of data to be proxies/streamed.

  • bufsize (Integer) (defaults to: 0)

    Buffer size to use

  • length (Integer) (defaults to: 0)

    Maximum number of bytes to proxy.

See Also:



1383
1384
1385
# File 'lib/eventmachine.rb', line 1383

def self.enable_proxy(from, to, bufsize=0, length=0)
  EM::start_proxy(from.signature, to.signature, bufsize, length)
end

.error_handler(cb = nil, &blk) ⇒ Object

Catch-all for errors raised during event loop callbacks.

Examples:


EventMachine.error_handler{ |e|
  puts "Error raised during event loop: #{e.message}"
}

Parameters:

  • cb (#call) (defaults to: nil)

    Global catch-all errback



1316
1317
1318
1319
1320
1321
1322
# File 'lib/eventmachine.rb', line 1316

def self.error_handler cb = nil, &blk
  if cb or blk
    @error_handler = cb || blk
  elsif instance_variable_defined? :@error_handler
    remove_instance_variable :@error_handler
  end
end

.fork_reactor(&block) ⇒ Object

Forks a new process, properly stops the reactor and then calls run inside of it again, passing your block.



257
258
259
260
261
262
263
264
265
266
267
268
269
270
# File 'lib/eventmachine.rb', line 257

def self.fork_reactor &block
  # This implementation is subject to change, especially if we clean up the relationship
  # of EM#run to @reactor_running.
  # Original patch by Aman Gupta.
  #
  Kernel.fork do
    if self.reactor_running?
      self.stop_event_loop
      self.release_machine
      @reactor_running = false
    end
    self.run block
  end
end

.get_max_timersInteger

Gets the current maximum number of allowed timers

Returns:

  • (Integer)

    Maximum number of timers that may be outstanding at any given time



907
908
909
# File 'lib/eventmachine.rb', line 907

def self.get_max_timers
  get_max_timer_count
end

.heartbeat_intervalInteger

Retrieve the heartbeat interval. This is how often EventMachine will check for dead connections that have had an inactivity timeout set via EventMachine::Connection#set_comm_inactivity_timeout. Default is 2 seconds.

Returns:

  • (Integer)

    Heartbeat interval, in seconds



1402
1403
1404
# File 'lib/eventmachine.rb', line 1402

def self.heartbeat_interval
  EM::get_heartbeat_interval
end

.heartbeat_interval=(time) ⇒ Object

Set the heartbeat interval. This is how often EventMachine will check for dead connections that have had an inactivity timeout set via EventMachine::Connection#set_comm_inactivity_timeout. Takes a Numeric number of seconds. Default is 2.

Parameters:

  • time (Integer)

    Heartbeat interval, in seconds



1411
1412
1413
# File 'lib/eventmachine.rb', line 1411

def self.heartbeat_interval=(time)
  EM::set_heartbeat_interval time.to_f
end

.next_tick(pr = nil, &block) ⇒ Object

Schedules a proc for execution immediately after the next "turn" through the reactor core. An advanced technique, this can be useful for improving memory management and/or application responsiveness, especially when scheduling large amounts of data for writing to a network connection.

This method takes either a single argument (which must be a callable object) or a block.

Parameters:

  • pr (#call) (defaults to: nil)

    A callable object to run

Raises:

  • (ArgumentError)


1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
# File 'lib/eventmachine.rb', line 1079

def self.next_tick pr=nil, &block
  # This works by adding to the @resultqueue that's used for #defer.
  # The general idea is that next_tick is used when we want to give the reactor a chance
  # to let other operations run, either to balance the load out more evenly, or to let
  # outbound network buffers drain, or both. So we probably do NOT want to block, and
  # we probably do NOT want to be spinning any threads. A program that uses next_tick
  # but not #defer shouldn't suffer the penalty of having Ruby threads running. They're
  # extremely expensive even if they're just sleeping.

  raise ArgumentError, "no proc or block given" unless ((pr && pr.respond_to?(:call)) or block)
  @next_tick_mutex.synchronize do
    @next_tick_queue << ( pr || block )
  end
  signal_loopbreak if reactor_running?
end

.open_datagram_socket(address, port, handler = nil, *args) ⇒ Object

Used for UDP-based protocols. Its usage is similar to that of start_server.

This method will create a new UDP (datagram) socket and bind it to the address and port that you specify. The normal callbacks (see start_server) will be called as events of interest occur on the newly-created socket, but there are some differences in how they behave.

EventMachine::Connection#receive_data will be called when a datagram packet is received on the socket, but unlike TCP sockets, the message boundaries of the received data will be respected. In other words, if the remote peer sent you a datagram of a particular size, you may rely on EventMachine::Connection#receive_data to give you the exact data in the packet, with the original data length. Also observe that Connection#receive_data may be called with a zero-length data payload, since empty datagrams are permitted in UDP.

EventMachine::Connection#send_data is available with UDP packets as with TCP, but there is an important difference. Because UDP communications are connectionless, there is no implicit recipient for the packets you send. Ordinarily you must specify the recipient for each packet you send. However, EventMachine provides for the typical pattern of receiving a UDP datagram from a remote peer, performing some operation, and then sending one or more packets in response to the same remote peer. To support this model easily, just use EventMachine::Connection#send_data in the code that you supply for EventMachine::Connection#receive_data.

EventMachine will provide an implicit return address for any messages sent to EventMachine::Connection#send_data within the context of a EventMachine::Connection#receive_data callback, and your response will automatically go to the correct remote peer.

Observe that the port number that you supply to open_datagram_socket may be zero. In this case, EventMachine will create a UDP socket that is bound to an ephemeral port. This is not appropriate for servers that must publish a well-known port to which remote peers may send datagrams. But it can be useful for clients that send datagrams to other servers. If you do this, you will receive any responses from the remote servers through the normal EventMachine::Connection#receive_data callback. Observe that you will probably have issues with firewalls blocking the ephemeral port numbers, so this technique is most appropriate for LANs.

If you wish to send datagrams to arbitrary remote peers (not necessarily ones that have sent data to which you are responding), then see EventMachine::Connection#send_datagram.

DO NOT call send_data from a datagram socket outside of a EventMachine::Connection#receive_data method. Use EventMachine::Connection#send_datagram. If you do use EventMachine::Connection#send_data outside of a EventMachine::Connection#receive_data method, you'll get a confusing error because there is no "peer," as #send_data requires (inside of EventMachine::Connection#receive_data, EventMachine::Connection#send_data "fakes" the peer as described above).

Parameters:

  • address (String)

    IP address

  • port (String)

    Port

  • handler (Class, Module) (defaults to: nil)

    A class or a module that implements connection lifecycle callbacks.



855
856
857
858
859
860
861
862
863
864
865
866
# File 'lib/eventmachine.rb', line 855

def self.open_datagram_socket address, port, handler=nil, *args
  # Replaced the implementation on 01Oct06. Thanks to Tobias Gustafsson for pointing
  # out that this originally did not take a class but only a module.


  klass = klass_from_handler(Connection, handler, *args)
  s = open_udp_socket address, port.to_i
  c = klass.new s, *args
  @conns[s] = c
  block_given? and yield c
  c
end

.popen(cmd, handler = nil, *args) {|c| ... } ⇒ Object

Note:

This method is not supported on Microsoft Windows

Runs an external process.

Examples:


module RubyCounter
  def post_init
    # count up to 5
    send_data "5\n"
  end
  def receive_data data
    puts "ruby sent me: #{data}"
  end
  def unbind
    puts "ruby died with exit status: #{get_status.exitstatus}"
  end
end

EventMachine.run {
  EventMachine.popen("ruby -e' $stdout.sync = true; gets.to_i.times{ |i| puts i+1; sleep 1 } '", RubyCounter)
}

Yields:

  • (c)

See Also:



1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
# File 'lib/eventmachine.rb', line 1156

def self.popen cmd, handler=nil, *args
  # At this moment, it's only available on Unix.
  # Perhaps misnamed since the underlying function uses socketpair and is full-duplex.

  klass = klass_from_handler(Connection, handler, *args)
  w = Shellwords::shellwords( cmd )
  w.unshift( w.first ) if w.first
  s = invoke_popen( w )
  c = klass.new s, *args
  @conns[s] = c
  yield(c) if block_given?
  c
end

.reactor_running?Boolean

Tells you whether the EventMachine reactor loop is currently running.

Useful when writing libraries that want to run event-driven code, but may be running in programs that are already event-driven. In such cases, if reactor_running? returns false, your code can invoke run and run your application code inside the block passed to that method. If this method returns true, just execute your event-aware code.

Returns:

  • (Boolean)

    true if the EventMachine reactor loop is currently running



1180
1181
1182
# File 'lib/eventmachine.rb', line 1180

def self.reactor_running?
  (@reactor_running || false)
end

.reactor_thread?Boolean

Returns true if the calling thread is the same thread as the reactor.

Returns:

  • (Boolean)

    true if the calling thread is the same thread as the reactor.



241
242
243
# File 'lib/eventmachine.rb', line 241

def self.reactor_thread?
  Thread.current == @reactor_thread
end

.reconnect(server, port, handler) ⇒ Object

Connect to a given host/port and re-use the provided Connection instance. Consider also EventMachine::Connection#reconnect.



764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
# File 'lib/eventmachine.rb', line 764

def self.reconnect server, port, handler
  # Observe, the test for already-connected FAILS if we call a reconnect inside post_init,
  # because we haven't set up the connection in @conns by that point.
  # RESIST THE TEMPTATION to "fix" this problem by redefining the behavior of post_init.
  #
  # Changed 22Nov06: if called on an already-connected handler, just return the
  # handler and do nothing more. Originally this condition raised an exception.
  # We may want to change it yet again and call the block, if any.

  raise "invalid handler" unless handler.respond_to?(:connection_completed)
  #raise "still connected" if @conns.has_key?(handler.signature)
  return handler if @conns.has_key?(handler.signature)

  s = if port
        connect_server server, port
      else
        connect_unix_server server
      end
  handler.signature = s
  @conns[s] = handler
  block_given? and yield handler
  handler
end

.run(blk = nil, tail = nil, &block) ⇒ Object

Note:

This method blocks calling thread. If you need to start EventMachine event loop from a Web app running on a non event-driven server (Unicorn, Apache Passenger, Mongrel), do it in a separate thread like demonstrated in one of the examples.

Initializes and runs an event loop. This method only returns if code inside the block passed to this method calls stop_event_loop. The block is executed after initializing its internal event loop but before running the loop, therefore this block is the right place to call any code that needs event loop to run, for example, start_server, connect or similar methods of libraries that use EventMachine under the hood (like EventMachine::HttpRequest.new or AMQP.start).

Programs that are run for long periods of time (e.g. servers) usually start event loop by calling run, and let it run "forever". It's also possible to use run to make a single client-connection to a remote server, process the data flow from that single connection, and then call stop_event_loop to stop, in other words, to run event loop for a short period of time (necessary to complete some operation) and then shut it down.

Once event loop is running, it is perfectly possible to start multiple servers and clients simultaneously: content-aware proxies like Proxymachine do just that.

Using EventMachine with Ruby on Rails and other Web application frameworks

Standalone applications often run event loop on the main thread, thus blocking for their entire lifespan. In case of Web applications, if you are running an EventMachine-based app server such as Thin or Goliath, they start event loop for you. Servers like Unicorn, Apache Passenger or Mongrel occupy main Ruby thread to serve HTTP(S) requests. This means that calling run on the same thread is not an option (it will result in Web server never binding to the socket). In that case, start event loop in a separate thread as demonstrated below.

Examples:

Starting EventMachine event loop in the current thread to run the "Hello, world"-like Echo server example


#!/usr/bin/env ruby

require 'rubygems' # or use Bundler.setup
require 'eventmachine'

class EchoServer < EM::Connection
  def receive_data(data)
    send_data(data)
  end
end

EventMachine.run do
  EventMachine.start_server("0.0.0.0", 10000, EchoServer)
end

Starting EventMachine event loop in a separate thread


# doesn't block current thread, can be used with Ruby on Rails, Sinatra, Merb, Rack
# and any other application server that occupies main Ruby thread.
Thread.new { EventMachine.run }

See Also:



151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
# File 'lib/eventmachine.rb', line 151

def self.run blk=nil, tail=nil, &block
  # Obsoleted the use_threads mechanism.
  # 25Nov06: Added the begin/ensure block. We need to be sure that release_machine
  # gets called even if an exception gets thrown within any of the user code
  # that the event loop runs. The best way to see this is to run a unit
  # test with two functions, each of which calls {EventMachine.run} and each of
  # which throws something inside of #run. Without the ensure, the second test
  # will start without release_machine being called and will immediately throw

  #
  if reactor_running? and @reactor_pid != Process.pid
    # Reactor was started in a different parent, meaning we have forked.
    # Clean up reactor state so a new reactor boots up in this child.
    stop_event_loop
    release_machine
    @reactor_running = false
  end

  tail and @tails.unshift(tail)

  if reactor_running?
    (b = blk || block) and b.call # next_tick(b)
  else
    @conns = {}
    @acceptors = {}
    @timers = {}
    @wrapped_exception = nil
    @next_tick_queue ||= []
    @tails ||= []
    begin
      @reactor_pid = Process.pid
      @reactor_running = true
      initialize_event_machine
      (b = blk || block) and add_timer(0, b)
      if @next_tick_queue && !@next_tick_queue.empty?
        add_timer(0) { signal_loopbreak }
      end
      @reactor_thread = Thread.current
      run_machine
    ensure
      until @tails.empty?
        @tails.pop.call
      end

      begin
        release_machine
      ensure
        if @threadpool
          @threadpool.each { |t| t.exit }
          @threadpool.each do |t|
            next unless t.alive?
            begin
              # Thread#kill! does not exist on 1.9 or rbx, and raises
              # NotImplemented on jruby
              t.kill!
            rescue NoMethodError, NotImplementedError
              t.kill
              # XXX t.join here?
            end
          end
          @threadqueue = nil
          @resultqueue = nil
          @threadpool = nil
          @all_threads_spawned = false
        end

        @next_tick_queue = []
      end
      @reactor_running = false
      @reactor_thread = nil
    end

    raise @wrapped_exception if @wrapped_exception
  end
end

.run_block(&block) ⇒ Object

Sugars a common use case. Will pass the given block to #run, but will terminate the reactor loop and exit the function as soon as the code in the block completes. (Normally, run keeps running indefinitely, even after the block supplied to it finishes running, until user code calls stop)



232
233
234
235
236
237
238
# File 'lib/eventmachine.rb', line 232

def self.run_block &block
  pr = proc {
    block.call
    EventMachine::stop
  }
  run(&pr)
end

.schedule(*a, &b) ⇒ Object

Runs the given callback on the reactor thread, or immediately if called from the reactor thread. Accepts the same arguments as Callback



247
248
249
250
251
252
253
254
# File 'lib/eventmachine.rb', line 247

def self.schedule(*a, &b)
  cb = Callback(*a, &b)
  if reactor_running? && reactor_thread?
    cb.call
  else
    next_tick { cb.call }
  end
end

.set_descriptor_table_size(n_descriptors = nil) ⇒ Integer

Sets the maximum number of file or socket descriptors that your process may open. If you call this method with no arguments, it will simply return the current size of the descriptor table without attempting to change it.

The new limit on open descriptors only applies to sockets and other descriptors that belong to EventMachine. It has no effect on the number of descriptors you can create in ordinary Ruby code.

Not available on all platforms. Increasing the number of descriptors beyond its default limit usually requires superuser privileges. (See set_effective_user for a way to drop superuser privileges while your program is running.)

Parameters:

  • n_descriptors (Integer) (defaults to: nil)

    The maximum number of file or socket descriptors that your process may open

Returns:

  • (Integer)

    The new descriptor table size.



1126
1127
1128
# File 'lib/eventmachine.rb', line 1126

def self.set_descriptor_table_size n_descriptors=nil
  EventMachine::set_rlimit_nofile n_descriptors
end

.set_effective_user(username) ⇒ Object

Note:

This method has no effective implementation on Windows or in the pure-Ruby implementation of EventMachine

A wrapper over the setuid system call. Particularly useful when opening a network server on a privileged port because you can use this call to drop privileges after opening the port. Also very useful after a call to set_descriptor_table_size, which generally requires that you start your process with root privileges.

This method is intended for use in enforcing security requirements, consequently it will throw a fatal error and end your program if it fails.

Parameters:

  • username (String)

    The effective name of the user whose privilege-level your process should attain.



1107
1108
1109
# File 'lib/eventmachine.rb', line 1107

def self.set_effective_user username
  EventMachine::setuid_string username
end

.set_max_timers(ct) ⇒ Object

Note:

This method has to be used before event loop is started.

Sets the maximum number of timers and periodic timers that may be outstanding at any given time. You only need to call set_max_timers if you need more than the default number of timers, which on most platforms is 1000.

Parameters:

  • ct (Integer)

    Maximum number of timers that may be outstanding at any given time

See Also:



900
901
902
# File 'lib/eventmachine.rb', line 900

def self.set_max_timers ct
  set_max_timer_count ct
end

.set_quantum(mills) ⇒ Object

For advanced users. This function sets the default timer granularity, which by default is slightly smaller than 100 milliseconds. Call this function to set a higher or lower granularity. The function affects the behavior of add_timer and add_periodic_timer. Most applications will not need to call this function.

Avoid setting the quantum to very low values because that may reduce performance under some extreme conditions. We recommend that you not use values lower than 10.

This method only can be used if event loop is running.

Parameters:

  • mills (Integer)

    New timer granularity, in milliseconds

See Also:



885
886
887
# File 'lib/eventmachine.rb', line 885

def self.set_quantum mills
  set_timer_quantum mills.to_i
end

.spawn(&block) ⇒ Object

Spawn an erlang-style process



69
70
71
72
73
# File 'lib/em/spawnable.rb', line 69

def self.spawn &block
  s = SpawnedProcess.new
  s.set_receiver block
  s
end

.start_server(server, port = nil, handler = nil, *args, &block) ⇒ Object

Note:

Don't forget that in order to bind to ports < 1024 on Linux, *BSD and Mac OS X your process must have superuser privileges.

Initiates a TCP server (socket acceptor) on the specified IP address and port.

The IP address must be valid on the machine where the program runs, and the process must be privileged enough to listen on the specified port (on Unix-like systems, superuser privileges are usually required to listen on any port lower than 1024). Only one listener may be running on any given address/port combination. start_server will fail if the given address and port are already listening on the machine, either because of a prior call to start_server or some unrelated process running on the machine. If start_server succeeds, the new network listener becomes active immediately and starts accepting connections from remote peers, and these connections generate callback events that are processed by the code specified in the handler parameter to start_server.

The optional handler which is passed to this method is the key to EventMachine's ability to handle particular network protocols. The handler parameter passed to start_server must be a Ruby Module that you must define. When the network server that is started by start_server accepts a new connection, it instantiates a new object of an anonymous class that is inherited from Connection, into which your handler module have been included. Arguments passed into start_server after the class name are passed into the constructor during the instantiation.

Your handler module may override any of the methods in Connection, such as EventMachine::Connection#receive_data, in order to implement the specific behavior of the network protocol.

Callbacks invoked in response to network events always take place within the execution context of the object derived from Connection extended by your handler module. There is one object per connection, and all of the callbacks invoked for a particular connection take the form of instance methods called against the corresponding Connection object. Therefore, you are free to define whatever instance variables you wish, in order to contain the per-connection state required by the network protocol you are implementing.

start_server is usually called inside the block passed to run, but it can be called from any EventMachine callback. start_server will fail unless the EventMachine event loop is currently running (which is why it's often called in the block suppled to run).

You may call start_server any number of times to start up network listeners on different address/port combinations. The servers will all run simultaneously. More interestingly, each individual call to start_server can specify a different handler module and thus implement a different network protocol from all the others.

Examples:


require 'rubygems'
require 'eventmachine'

# Here is an example of a server that counts lines of input from the remote
# peer and sends back the total number of lines received, after each line.
# Try the example with more than one client connection opened via telnet,
# and you will see that the line count increments independently on each
# of the client connections. Also very important to note, is that the
# handler for the receive_data function, which our handler redefines, may
# not assume that the data it receives observes any kind of message boundaries.
# Also, to use this example, be sure to change the server and port parameters
# to the start_server call to values appropriate for your environment.
module LineCounter
  MaxLinesPerConnection = 10

  def post_init
    puts "Received a new connection"
    @data_received = ""
    @line_count = 0
  end

  def receive_data data
    @data_received << data
    while @data_received.slice!( /^[^\n]*[\n]/m )
      @line_count += 1
      send_data "received #{@line_count} lines so far\r\n"
      @line_count == MaxLinesPerConnection and close_connection_after_writing
    end
  end
end

EventMachine.run {
  host, port = "192.168.0.100", 8090
  EventMachine.start_server host, port, LineCounter
  puts "Now accepting connections on address #{host}, port #{port}..."
  EventMachine.add_periodic_timer(10) { $stderr.write "*" }
}

Parameters:

  • server (String)

    Host to bind to.

  • port (Integer) (defaults to: nil)

    Port to bind to.

  • handler (Module, Class) (defaults to: nil)

    A module or class that implements connection callbacks

See Also:



514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
# File 'lib/eventmachine.rb', line 514

def self.start_server server, port=nil, handler=nil, *args, &block
  begin
    port = Integer(port)
  rescue ArgumentError, TypeError
    # there was no port, so server must be a unix domain socket
    # the port argument is actually the handler, and the handler is one of the args
    args.unshift handler if handler
    handler = port
    port = nil
  end if port

  klass = klass_from_handler(Connection, handler, *args)

  s = if port
        start_tcp_server server, port
      else
        start_unix_server server
      end
  @acceptors[s] = [klass,args,block]
  s
end

.start_unix_domain_server(filename, *args, &block) ⇒ Object

Start a Unix-domain server.

Note that this is an alias for start_server, which can be used to start both TCP and Unix-domain servers.

See Also:



549
550
551
# File 'lib/eventmachine.rb', line 549

def self.start_unix_domain_server filename, *args, &block
  start_server filename, *args, &block
end

.stop_event_loopObject

Causes the processing loop to stop executing, which will cause all open connections and accepting servers to be run down and closed. Connection termination callbacks added using add_shutdown_hook will be called as part of running this method.

When all of this processing is complete, the call to run which started the processing loop will return and program flow will resume from the statement following run call.

Examples:

Stopping a running EventMachine event loop


require 'rubygems'
require 'eventmachine'

module Redmond
  def post_init
    puts "We're sending a dumb HTTP request to the remote peer."
    send_data "GET / HTTP/1.1\r\nHost: www.microsoft.com\r\n\r\n"
  end

  def receive_data data
    puts "We received #{data.length} bytes from the remote peer."
    puts "We're going to stop the event loop now."
    EventMachine::stop_event_loop
  end

  def unbind
    puts "A connection has terminated."
  end
end

puts "We're starting the event loop now."
EventMachine.run {
  EventMachine.connect "www.microsoft.com", 80, Redmond
}
puts "The event loop has stopped."

# This program will produce approximately the following output:
#
# We're starting the event loop now.
# We're sending a dumb HTTP request to the remote peer.
# We received 1440 bytes from the remote peer.
# We're going to stop the event loop now.
# A connection has terminated.
# The event loop has stopped.


414
415
416
# File 'lib/eventmachine.rb', line 414

def self.stop_event_loop
  EventMachine::stop
end

.stop_server(signature) ⇒ Object

Stop a TCP server socket that was started with start_server.

See Also:



539
540
541
# File 'lib/eventmachine.rb', line 539

def self.stop_server signature
  EventMachine::stop_tcp_server signature
end

.system(cmd, *args, &cb) ⇒ Object

EM::system is a simple wrapper for EM::popen. It is similar to Kernel::system, but requires a single string argument for the command and performs no shell expansion.

The block or proc passed to EM::system is called with two arguments: the output generated by the command, and a Process::Status that contains information about the command's execution.

EM.run{ EM.system('ls'){ |output,status| puts output if status.exitstatus == 0 } }

You can also supply an additional proc to send some data to the process:

EM.run{ EM.system('sh', proc{ |process| process.send_data("echo hello\n") process.send_data("exit\n") }, proc{ |out,status| puts(out) }) }

Like EventMachine.popen, EventMachine.system currently does not work on windows. It returns the pid of the spawned process.



112
113
114
115
116
117
118
119
120
121
122
# File 'lib/em/processes.rb', line 112

def EventMachine::system cmd, *args, &cb
  cb ||= args.pop if args.last.is_a? Proc
  init = args.pop if args.last.is_a? Proc

  # merge remaining arguments into the command
  cmd = ([cmd] + args.map{|a|a.to_s.dump}).join(' ')

  EM.get_subprocess_pid(EM.popen(cmd, SystemCmd, cb) do |c|
    init[c] if init
  end.signature)
end

.tick_loop(*a, &b) ⇒ Object

Creates and immediately starts an EventMachine::TickLoop



3
4
5
# File 'lib/em/tick_loop.rb', line 3

def self.tick_loop(*a, &b)
  TickLoop.new(*a, &b).start
end

.watch(io, handler = nil, *args, &blk) ⇒ Object

watch registers a given file descriptor or IO object with the eventloop. The file descriptor will not be modified (it will remain blocking or non-blocking).

The eventloop can be used to process readable and writable events on the file descriptor, using EventMachine::Connection#notify_readable= and EventMachine::Connection#notify_writable=

EventMachine::Connection#notify_readable? and EventMachine::Connection#notify_writable? can be used to check what events are enabled on the connection.

To detach the file descriptor, use EventMachine::Connection#detach

Examples:


module SimpleHttpClient
  def notify_readable
    header = @io.readline

    if header == "\r\n"
      # detach returns the file descriptor number (fd == @io.fileno)
      fd = detach
    end
  rescue EOFError
    detach
  end

  def unbind
    EM.next_tick do
      # socket is detached from the eventloop, but still open
      data = @io.read
    end
  end
end

EventMachine.run {
  sock = TCPSocket.new('site.com', 80)
  sock.write("GET / HTTP/1.0\r\n\r\n")
  conn = EventMachine.watch(sock, SimpleHttpClient)
  conn.notify_readable = true
}

Author:

  • Riham Aldakkak (eSpace Technologies)



719
720
721
# File 'lib/eventmachine.rb', line 719

def EventMachine::watch io, handler=nil, *args, &blk
  attach_io io, true, handler, *args, &blk
end

.watch_file(filename, handler = nil, *args) ⇒ Object

Note:

The ability to pick up on the new filename after a rename is not yet supported. Calling #path will always return the filename you originally used.

EventMachine's file monitoring API. Currently supported are the following events on individual files, using inotify on Linux systems, and kqueue for *BSD and Mac OS X:

  • File modified (written to)
  • File moved/renamed
  • File deleted

EventMachine::watch_file takes a filename and a handler Module containing your custom callback methods. This will setup the low level monitoring on the specified file, and create a new EventMachine::FileWatch object with your Module mixed in. FileWatch is a subclass of Connection, so callbacks on this object work in the familiar way. The callbacks that will be fired by EventMachine are:

  • file_modified
  • file_moved
  • file_deleted

You can access the filename being monitored from within this object using EventMachine::FileWatch#path.

When a file is deleted, EventMachine::FileWatch#stop_watching will be called after your file_deleted callback, to clean up the underlying monitoring and remove EventMachine's reference to the now-useless FileWatch instance. This will in turn call unbind, if you wish to use it.

The corresponding system-level Errno will be raised when attempting to monitor non-existent files, files with wrong permissions, or if an error occurs dealing with inotify/kqueue.

Examples:


# Before running this example, make sure we have a file to monitor:
# $ echo "bar" > /tmp/foo

module Handler
  def file_modified
    puts "#{path} modified"
  end

  def file_moved
    puts "#{path} moved"
  end

  def file_deleted
    puts "#{path} deleted"
  end

  def unbind
    puts "#{path} monitoring ceased"
  end
end

# for efficient file watching, use kqueue on Mac OS X
EventMachine.kqueue = true if EventMachine.kqueue?

EventMachine.run {
  EventMachine.watch_file("/tmp/foo", Handler)
}

# $ echo "baz" >> /tmp/foo    =>    "/tmp/foo modified"
# $ mv /tmp/foo /tmp/oof      =>    "/tmp/foo moved"
# $ rm /tmp/oof               =>    "/tmp/foo deleted"

Parameters:

  • filename (String)

    Local path to the file to watch.

  • handler (Class, Module) (defaults to: nil)

    A class or module that implements event handlers associated with the file.



1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
# File 'lib/eventmachine.rb', line 1262

def self.watch_file(filename, handler=nil, *args)
  klass = klass_from_handler(FileWatch, handler, *args)

  s = EM::watch_filename(filename)
  c = klass.new s, *args
  # we have to set the path like this because of how Connection.new works
  c.instance_variable_set("@path", filename)
  @conns[s] = c
  block_given? and yield c
  c
end

.watch_process(pid, handler = nil, *args) ⇒ Object

EventMachine's process monitoring API. On Mac OS X and *BSD this method is implemented using kqueue.

Examples:


module ProcessWatcher
  def process_exited
    put 'the forked child died!'
  end
end

pid = fork{ sleep }

EventMachine.run {
  EventMachine.watch_process(pid, ProcessWatcher)
  EventMachine.add_timer(1){ Process.kill('TERM', pid) }
}

Parameters:

  • pid (Integer)

    PID of the process to watch.

  • handler (Class, Module) (defaults to: nil)

    A class or module that implements event handlers associated with the file.



1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
# File 'lib/eventmachine.rb', line 1293

def self.watch_process(pid, handler=nil, *args)
  pid = pid.to_i

  klass = klass_from_handler(ProcessWatch, handler, *args)

  s = EM::watch_pid(pid)
  c = klass.new s, *args
  # we have to set the path like this because of how Connection.new works
  c.instance_variable_set("@pid", pid)
  @conns[s] = c
  block_given? and yield c
  c
end