Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Actors11-28-2011Opening DiscussionShould you use XML or direct access for your project?Finishing the file example.Challenges in ParallelismWe have already seen that you need to have your programs multithreaded to take full advantage of modern processors.Unfortunately, the standard threading model can be very difficult to use. Race-conditions and deadlock can be difficult to avoid and are extremely challenging to debug.ActorsAnother approach to parallelism is the actor model. This is the main approach to programming Erlang which is used a lot in telecom.Scala includes an actor package that supports this model.With actors, data isn't shared. You send messages instead of calling methods.Let's take a minute to look in the API.MessagesActors respond to messages that are sent to them. The actions of actors are done in parallel.Use ! to send a message.It is ideal to use case classes for messages. That way you get pattern matching and immutable data.Messages go to an actors inbox. Sending messages does not block.The act MethodThe main logic for an actor goes in the act method.Call receive to process a message.Put the receive inside a loop to do so repeatedly.Receive takes a partial function that handles the appropriate messages.Don't block.Short VersionThere is a helper method in the Actor object that takes a function for the body of the act method.You should import scala.actors.Actor, then you can call actor { body }.Playing with ActorsLet's play a little with Actors given what we know so far.Minute
View Full Document
Unlocking...