Akka: partial function chaining and sender for beh

I try to implement different behaviours using Akka actors. In my approach, traits define message handling behaviours. The concrete actors mix in those traits and then pick which behaviours they want when building out their receive function by using partial function chaining.

Unfortunately, it seems there is some trouble considering the instanciation of the sender of messages. As shown in the following console message alice is unable to indetify the sender of the message "Good" which is bob.

alice send Fine? to bob bob replies Good to Actor[akka://StrategiesSystem/user/alice] alice receives Good from Actor[akka://StrategiesSystem/deadLetters]

As you will see in my code the expected result is that bob should be stopped which is not the case.

Your help will be aprreciated.

Thanks in advance,

Max.

import akka.actor._
import scala.util.Random

//The messages which can be exchanged
sealed trait Move
case object StartMessage extends Move
case object StopMessage extends Move
case object Fine extends Move
case object Good extends Move
case object Bad extends Move

//The general class representing my actors
abstract class MyActor(val others: List[String]) extends Actor{ 
    //its name
    val name=self.path.name

    //it knows to choose on interlocutor
    val interlocteur=Random.shuffle(others.filterNot(p=> p==name)).head

    //All the actors are able to interpret the start/stop messages
    def metaReceive : Receive= {
        case StartMessage =>//start to ask question
            println(name+" send Fine? to "+interlocteur)
            context.actorSelection("../"+interlocteur) ! Fine
        case StopMessage =>
            println(name+" stops")
            context.stop(self)
  }
}

//An optimistic actor says it is fine
trait Optimistic{
  self: MyActor =>
  def handleFine:Receive = {
    case Fine => 
        println(self.name+" replies Good to "+sender)
        sender ! Good   
  }  
}

//A pessimistic actor says it is not fine
trait Pessimistic{
  self: MyActor =>
  def handleFine:Receive = {
    case Fine => 
        println(self.name+" replies Bad to "+sender)
        sender ! Bad
  }  
}

//An interpretor is an actor which is able to understand the reply
trait Interpretor{
  self: MyActor =>
  def handleAnswer:Receive = {
        case Good =>
            println(name+" receives Good from "+sender)
            sender ! StopMessage
        case Bad =>
            println(name+" receives Bad from "+sender)
            sender ! StopMessage
  }  
}

//My basic classes
class MyOptimisticActor(others: List[String]) extends  MyActor(others) with Optimistic{
  override def receive = metaReceive orElse handleFine //orElse ...
}

class MyPessimisticActor(others: List[String]) extends  MyActor(others) with Pessimistic{
  override def receive = metaReceive orElse handleFine //orElse ...
}
class MyInterpretorActor(others: List[String]) extends  MyActor(others) with Interpretor{
  override def receive = metaReceive orElse handleAnswer
}

//My application
object TestStrategies extends Application {
  val system = ActorSystem("StrategiesSystem")
  val names= List("alice","bob","carla")
  val alice = system.actorOf(Props(new MyInterpretorActor(names)), name = "alice")// alice is able to ask question and interpret answer
  val bob = system.actorOf(Props(new MyOptimisticActor(names)), name = "bob") // bob is able to reply and it is fine
  val carla = system.actorOf(Props(new MyPessimisticActor(names)), name = "carla") //carla is able to reply and it is not fine
  alice ! StartMessage //alice must ask a question
}

Don't use