observer

Observer pattern

Define a one to many dependency between objects so that when one object changes state, all its dependents are notified and updated automatically.

Structure

classDiagram
class Observer {
    Update()
}

class ConcreteObserver {
    Update()
}

class Subject {
    Attach(Observer o)
    Detach(Observer o)
    Notify()
}

class ConcreteSubject {
    GetState()
    SetState()
}

ConcreteSubject--|>Subject : implements
ConcreteObserver--|>Observer : implements
Subject-->Observer : has-many

Loading

Notice that a Subject can have many Observer's, each Observer has a ConcreteObserver implementation, that knows how to Update() themselves whenever the Subject notifies (Notify()) them.

How-To

The implementation itself may vary depending on how much data we want to share between the two parts, but the top-level idea remains the same:

• We start by defining two interfaces, one for the Subject and one for the Observers:

interface Subject
{
  void Attach(Observer o);
  void Detach(Observer o);
  void Notify();
}

interface Observer
{
  void Update(State s);
}

• We can then build concrete classes implementing these interfaces accordingly. First, the Subject:

class ConcreteSubject : Subject
{
  private SubjectState state { get; set; };
  private List<Observer> observers { get; set; }

  public ConcreteSubject()
  {
    this.observers = [];
  }

  public void Attach(Observer o)
  {
    this.observers.Add(o);
  }

  public void Detach(Observer o)
  {
    this.observers.Remove(o);
  }

  public void Notify()
  {
    this.observers.ForEach(o => o.Update(this.state));
  }
}

• Then, the Observer's:

class ConcreteObserver : Observer
{
  public ConcreteObserver()
  {}

  public void Update(State s)
  {
    // updates internal state accordingly
  }
}

GoF's book also describes a more complex way of handling state changes, should the state be too complicated to be handled by being passed down directory to Observer's, we could add a ChangeManager (using the Mediator Pattern) to abstract away the complex parts, therefore making the communication more streamlined.

Working example

This working example is extract from the Head First Design Patterns book and implements a weather application. This application is attached to an imaginary WeatherStation that is able to collect measurements and send them over to a WeatherData class, using the SetMeasurements() method. The WeatherData implements the Observer Pattern and it represents a Subject, so it's able to broadcast about its new updates to all the observers. For the observers we implemented Display's. Each display has a different logic to render a final text, but they all share the same interface. Each display class also implements the Observer interface to be able to receive the updates from WeatherData. I had to use generics in the Observer interface definition so we can pass a State object, to encapsulate the measurements fields. Make sure to check out the WeatherStation project for full implementation details.