Module: EventMachine::Deferrable

Included in:
DefaultDeferrable, DeferrableChildProcess, FileStreamer, Protocols::HttpClient, Protocols::HttpClient2::Request, Protocols::Memcache, Protocols::SmtpClient, Protocols::TcpConnectTester
Defined in:
lib/em/future.rb,
lib/em/deferrable.rb

Class Method Summary collapse

Instance Method Summary collapse

Class Method Details

.future(arg, cb = nil, eb = nil, &blk) ⇒ Object

A future is a sugaring of a typical deferrable usage. – Evaluate arg (which may be an expression or a block). What’s the class of arg? If arg is an ordinary expression, then return it. If arg is deferrable (responds to :set_deferred_status), then look at the arguments. If either callback or errback are defined, then use them. If neither are defined, then use the supplied block (if any) as the callback. Then return arg.



44
45
46
47
48
49
50
51
52
53
54
55
56
57
 
# File 'lib/em/future.rb', line 44

def self.future arg, cb=nil, eb=nil, &blk
  arg = arg.call if arg.respond_to?(:call)

  if arg.respond_to?(:set_deferred_status)
    if cb || eb
      arg.callback(&cb) if cb
      arg.errback(&eb) if eb
    else
      arg.callback(&blk) if blk
    end
  end

  arg
end

Instance Method Details

#callback(&block) ⇒ Object

Specify a block to be executed if and when the Deferrable object receives a status of :succeeded. See #set_deferred_status for more information.

Calling this method on a Deferrable object whose status is not yet known will cause the callback block to be stored on an internal list. If you call this method on a Deferrable whose status is :succeeded, the block will be executed immediately, receiving the parameters given to the prior #set_deferred_status call.

– If there is no status, add a callback to an internal list. If status is succeeded, execute the callback immediately. If status is failed, do nothing.



43
44
45
46
47
48
49
50
51
52
 
# File 'lib/em/deferrable.rb', line 43

def callback &block
  return unless block
  @deferred_status ||= :unknown
  if @deferred_status == :succeeded
    block.call(*@deferred_args)
  elsif @deferred_status != :failed
    @callbacks ||= []
    @callbacks.unshift block # << block
  end
end

#cancel_timeout ⇒ Object

Cancels an outstanding timeout if any. Undoes the action of #timeout.



161
162
163
164
165
166
167
 
# File 'lib/em/deferrable.rb', line 161

def cancel_timeout
  @deferred_timeout ||= nil
  if @deferred_timeout
    @deferred_timeout.cancel
    @deferred_timeout = nil
  end
end

#errback(&block) ⇒ Object

Specify a block to be executed if and when the Deferrable object receives a status of :failed. See #set_deferred_status for more information. – If there is no status, add an errback to an internal list. If status is failed, execute the errback immediately. If status is succeeded, do nothing.



61
62
63
64
65
66
67
68
69
70
 
# File 'lib/em/deferrable.rb', line 61

def errback &block
  return unless block
  @deferred_status ||= :unknown
  if @deferred_status == :failed
    block.call(*@deferred_args)
  elsif @deferred_status != :succeeded
    @errbacks ||= []
    @errbacks.unshift block # << block
  end
end

#fail(*args) ⇒ Object Also known as: set_deferred_failure

Sugar for set_deferred_status(:failed, …)



179
180
181
 
# File 'lib/em/deferrable.rb', line 179

def fail *args
  set_deferred_status :failed, *args
end

#set_deferred_status(status, *args) ⇒ Object

Sets the “disposition” (status) of the Deferrable object. See also the large set of sugarings for this method. Note that if you call this method without arguments, no arguments will be passed to the callback/errback. If the user has coded these with arguments, then the user code will throw an argument exception. Implementors of deferrable classes must document the arguments they will supply to user callbacks.

OBSERVE SOMETHING VERY SPECIAL here: you may call this method even on the INSIDE of a callback. This is very useful when a previously-registered callback wants to change the parameters that will be passed to subsequently-registered ones.

You may give either :succeeded or :failed as the status argument.

If you pass :succeeded, then all of the blocks passed to the object using the #callback method (if any) will be executed BEFORE the #set_deferred_status method returns. All of the blocks passed to the object using #errback will be discarded.

If you pass :failed, then all of the blocks passed to the object using the #errback method (if any) will be executed BEFORE the #set_deferred_status method returns. All of the blocks passed to the object using # callback will be discarded.

If you pass any arguments to #set_deferred_status in addition to the status argument, they will be passed as arguments to any callbacks or errbacks that are executed. It’s your responsibility to ensure that the argument lists specified in your callbacks and errbacks match the arguments given in calls to #set_deferred_status, otherwise Ruby will raise an ArgumentError.

– We’re shifting callbacks off and discarding them as we execute them. This is valid because by definition callbacks are executed no more than once. It also has the magic effect of permitting recursive calls, which means that a callback can call #set_deferred_status and change the parameters that will be sent to subsequent callbacks down the chain.

Changed @callbacks and @errbacks from push/shift to unshift/pop, per suggestion by Kirk Haines, to work around the memory leak bug that still exists in many Ruby versions.

Changed 15Sep07: after processing callbacks or errbacks, CLEAR the other set of handlers. This gets us a little closer to the behavior of Twisted’s “deferred,” which only allows status to be set once. Prior to making this change, it was possible to “succeed” a Deferrable (triggering its callbacks), and then immediately “fail” it, triggering its errbacks! That is clearly undesirable, but it’s just as undesirable to raise an exception is status is set more than once on a Deferrable. The latter behavior would invalidate the idiom of resetting arguments by setting status from within a callback or errback, but more seriously it would cause spurious errors if a Deferrable was timed out and then an attempt was made to succeed it. See the comments under the new method #timeout.



124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
 
# File 'lib/em/deferrable.rb', line 124

def set_deferred_status status, *args
  cancel_timeout
  @errbacks ||= nil
  @callbacks ||= nil
  @deferred_status = status
  @deferred_args = args
  case @deferred_status
  when :succeeded
    if @callbacks
      while cb = @callbacks.pop
        cb.call(*@deferred_args)
      end
    end
    @errbacks.clear if @errbacks
  when :failed
    if @errbacks
      while eb = @errbacks.pop
        eb.call(*@deferred_args)
      end
    end
    @callbacks.clear if @callbacks
  end
end

#succeed(*args) ⇒ Object Also known as: set_deferred_success

Sugar for set_deferred_status(:succeeded, …)



172
173
174
 
# File 'lib/em/deferrable.rb', line 172

def succeed *args
  set_deferred_status :succeeded, *args
end

#timeout(seconds) ⇒ Object

Setting a timeout on a Deferrable causes it to go into the failed state after the Timeout expires (passing no arguments to the object’s errbacks). Setting the status at any time prior to a call to the expiration of the timeout will cause the timer to be cancelled.



153
154
155
156
157
 
# File 'lib/em/deferrable.rb', line 153

def timeout seconds
  cancel_timeout
  me = self
  @deferred_timeout = EventMachine::Timer.new(seconds) {me.fail}
end