Using actors to build message-based concurrency

The concept of actors as a strategy for running concurrent tasks has recently gained new popularity (thanks to Scala, Erlang, the Akka Framework, and other programming languages). Originally proposed by Carl Hewitt in 1973, actors offer a programming model that inherently guarantees concurrent code when compared with the traditional approach based on shared memory. Actors are similar to Object-Oriented objects (they follow the rule of encapsulation), but they can only communicate by sending immutable messages asynchronously to each other. The internal state of an actor is not exposed and can only be accessed from the outside by sending a message to the actor and receiving a reply.

Due to the asynchronous nature of the message passing pattern, an actor must be active in order to receive a message. One way to make an object active, is to allocate a system thread to it. Unfortunately, threads are a finite resource, and this is not good news for...