Uses for the Java Void Reference Type?

There is a Java Void -- uppercase V-- reference type. The only situation I have ever seen it used is to parameterize Callables

final Callable<Void> callable = new Callable<Void>() {
            public Void call() {
                foobar();
                return null;
            }
        };

Are there any other uses for the Java Void reference type? Can it ever be assigned anything other than null? If yes, do you have examples?

Void has become convention for a generic argument that you are not interested in. There is no reason why you should use any other non-instantiable type, such as System.

It is also often used in for example Map values (although Collections.newSetFromMap uses Boolean as maps don't have to accept null values) and java.security.PrivilegedAction.

You can create instance of Void using reflections, but they are not useful for anything. Void is a way to indicate a generic method returns nothing.

Constructor<Void> constructor = Void.class.getDeclaredConstructor();
constructor.setAccessible(true);
Void v = constructor.newInstance();
System.out.println("I have a " + v);

prints something like

I have a java.lang.Void@75636731

Future<Void> works like charm. :)

Given that there are no public constructors, I would say it can't be assigned anything other than null. I've only used it as a placeholder for "I don't need to use this generic parameter," as your example shows.

It could also be used in reflection, from what its Javadoc says: >The Void class is an uninstantiable placeholder class to hold a reference to the Class object representing the Java keyword void.

All the primitive wrapper classes (Integer, Byte, Boolean, Double, etc.) contain a reference to the corresponding primitive class in a static TYPE field, for example:

Integer.TYPE == int.class
Byte.TYPE == byte.class
Boolean.TYPE == boolean.class
Double.TYPE == double.class

Void was initially created as somewhere to put a reference to the void type:

Void.TYPE == void.class

However, you don't really gain anything by using Void.TYPE. When you use void.class it's much clearer that you're doing something with the void type.

As an aside, the last time I tried it, http://www.beanshell.org/">BeanShell</a> didn't recognise void.class, so you have to use Void.TYPE there.

When you use the visitor pattern it can be cleaner to use Void instead of Object when you want to be sure that the return value will be null

Example

public interface LeavesVisitor<OUT>
{
   OUT visit(Leaf1 leaf);

   OUT visit(Leaf2 leaf);
}

When you will implement your visitor you can explicitly set OUT to be Void so that you know your visitor will always return null, instead of using Object

public class MyVoidVisitor implements LeavesVisitor<Void>
{
    Void visit(Leaf1 leaf){
        //...do what you want on your leaf
        return null;
    }
    
    Void visit(Leaf2 leaf){
        //...do what you want on your leaf
        return null;
    }
}

Before generics, it was created for the reflection API, to hold TYPE returned by Method.getReturnType() for a void method, corresponding to the other primitive type classes.

EDIT: From the JavaDoc of Void: "The Void class is an uninstantiable placeholder class to hold a reference to the Class object representing the Java keyword void". Prior to Generics, I am aware of no use other than reflection.

As you can't instantiate Void, you can use Apache commons Null object, so

Null aNullObject = ObjectUtils.Null;
Null noObjectHere = null;

in the first line, you have an object, so aNullObject != null holds, while in the second line there is no reference, so noObjectHere == null holds

To answer the poster's original question, the usage for this is to differentiate between "the nothing" and "nothing", which are completely different things.

PS: Say no to Null object pattern

Void was created to wrap its primitive void type. Every primitive type has its corresponding Reference type. Void is used to instantiate a generic class or use of a generic method, a generic argument which you are not interested in. Here is an example...

public void onNewRegistration() {
    newRegistrationService.createNewUser(view.getUsername(), view.getPassword(),
            view.getInitialAmount(), view.getCurrency(), new AsyncCallback<Void>() {
      @Override
      public void onFailure(Throwable caught) {
          
      }

      @Override
      public void onSuccess(Void result) {
        eventBus.fireEvent(new NewRegistrationSuccessEvent());
      }
    });
  } 

Here, as you can see, I don't want anything from the server that I am asking to create a new registration ,but public interface AsyncCallback<T> { .... } is a generic interface, so i provide Void since generics don't accept primitive types

It is also commonly used on Async-IO completion callbacks when you dont have the need for an Attachment object. In that case you specify null to the IO operation and implement CompletionHandler<Integer,Void>.

It may be rare case but once, I used Void in aspect classes.

This was an aspect which runs after methods that has an @Log annotation, and logs the method returned and some info if the method return type is not void.

 @AfterReturning(value = "@annotation(log)", 
       returning = "returnValue", 
       argNames = "joinPoint, log, returnValue"
      )
    public void afterReturning(final JoinPoint joinPoint, final Log log,
            final Object returnValue) {

            Class<?> returnType = ((MethodSignature) joinPoint.getSignature())
            .getReturnType();
           if (Void.class.isAssignableFrom (returnType)) ) {
            //Do some log
         }
}