Unity is just an IoC "container". Google StructureMap and try it out instead. A bit easier to grok, I think, when the IoC stuff is new to you.

Basically, if you understand IoC then you understand that what you're doing is inverting the control for when an object gets created.

Without IoC:

public class MyClass
{
   IMyService _myService; 

   public MyClass()
   {
      _myService = new SomeConcreteService();    
   }
}

With IoC container:

public class MyClass
{
   IMyService _myService; 

   public MyClass(IMyService myService)
   {
      _myService = myService;    
   }
}

Without IoC, your class that relies on the IMyService has to new-up a concrete version of the service to use. And that is bad for a number of reasons (you've coupled your class to a specific concrete version of the IMyService, you can't unit test it easily, you can't change it easily, etc.)

With an IoC container you "configure" the container to resolve those dependencies for you. So with a constructor-based injection scheme, you just pass the interface to the IMyService dependency into the constructor. When you create the MyClass with your container, your container will resolve the IMyService dependency for you.

Using StructureMap, configuring the container looks like this:

StructureMapConfiguration.ForRequestedType<MyClass>().TheDefaultIsConcreteType<MyClass>();
StructureMapConfiguration.ForRequestedType<IMyService>().TheDefaultIsConcreteType<SomeConcreteService>();

So what you've done is told the container, "When someone requests the IMyService, give them a copy of the SomeConcreteService." And you've also specified that when someone asks for a MyClass, they get a concrete MyClass.

That's all an IoC container really does. They can do more, but that's the thrust of it - they resolve dependencies for you, so you don't have to (and you don't have to use the "new" keyword throughout your code).

Final step: when you create your MyClass, you would do this:

var myClass = ObjectFactory.GetInstance<MyClass>();

Hope that helps. Feel free to e-mail me.


I just watched the 30 minute Unity Dependency Injection IoC Screencast by David Hayden and felt that was a good explaination with examples. Here is a snippet from the show notes:

The screencast shows several common usages of the Unity IoC, such as:

  • Creating Types Not In Container
  • Registering and Resolving TypeMappings
  • Registering and Resolving Named TypeMappings
  • Singletons, LifetimeManagers, and the ContainerControlledLifetimeManager
  • Registering Existing Instances
  • Injecting Dependencies into Existing Instances
  • Populating the UnityContainer via App.config / Web.config
  • Specifying Dependencies via Injection API as opposed to Dependency Attributes
  • Using Nested ( Parent-Child ) Containers

Unity is a library like many others that allows you to get an instance of a requested type without having to create it yourself. So given.

public interface ICalculator
{
    void Add(int a, int b);
}

public class Calculator : ICalculator
{
    public void Add(int a, int b)
    {
        return a + b;
    }
}

You would use a library like Unity to register Calculator to be returned when the type ICalculator is requested aka IoC (Inversion of Control) (this example is theoretical, not technically correct).

IoCLlibrary.Register<ICalculator>.Return<Calculator>();

So now when you want an instance of an ICalculator you just...

Calculator calc = IoCLibrary.Resolve<ICalculator>();

IoC libraries can usually be configured to either hold a singleton or create a new instance every time you resolve a type.

Now let's say you have a class that relies on an ICalculator to be present you could have..

public class BankingSystem
{
    public BankingSystem(ICalculator calc)
    {
        _calc = calc;
    }

    private ICalculator _calc;
}

And you can setup the library to inject a object into the constructor when it's created.

So DI or Dependency Injection means to inject any object another might require.


Unity is an IoC. The point of IoC is to abstract the wiring of dependencies between types outside of the types themselves. This has a couple of advantages. First of all, it is done centrally which means you don't have to change a lot of code when dependencies change (which may be the case for unit tests).

Furthermore, if the wiring is done using configuration data instead of code, you can actually rewire the dependencies after deployment and thus change the behavior of the application without changing the code.


MSDN has a Developer's Guide to Dependency Injection Using Unity that may be useful.

The Developer's Guide starts with the basics of what dependency injection is, and continues with examples of how to use Unity for dependency injection. As of the February 2014 the Developer's Guide covers Unity 3.0, which was released in April 2013.