Question

I watched “Google Web Toolkit Architecture: Best Practices for Architecting Your GWT App” and I’m having trouble figuring out how the server side is supposed to work.

Slide 21 shows:

/** The name Command is taken */
interface Action<T extends Response> { }

interface Response { }

interface ContactsService extends RemoteService {
  <T extends Response> T execute(Action<T> action);
}

interface ContactsServiceAsync {
  <T extends Response> void execute(Action<T> action,
      AsyncCallback<T> callback);
}

I thought that meant I might be able to create

public ResponseSubclass execute(ActionSubclass action) { ... }

and gwt would choose that method when it matches my exact parameters, but it doesn’t. At the moment i’m using:

if (action.getClass().getName() == ActionSubclass.class.getName())
{
    return (T) execute((ActionSubclass)action);
}

but that means I have to keep adding ifs to that method every time I add an action and I have to use unchecked casts. Is there a better way to make this work?

note: from what I have read somewhere else, the command pattern would usually include the actions to be taken in the Ac subclass, but because this is passing a client object for the server take some action on, the execution of the action has to be separated.

Answer 1

Checkout GWT-Dispatch and how to use it here.

As an aside, this may be one of the best articles that references Ray Ryan’s IO talk and builds a Hello World app. The application uses GWT-Dispatch.

-JP