In an application, I have a stateless sessionbean as the facade of my subsystem. And I only need to let other sub-systems to call methods in the remote interface. In this case, the home interface contains only a "create" method with no parameter.
So, why should we get the stateless sessionbean through creating method in its home object instead of retrieving a remote object directly from JNDI?
If using JNDI can we get the bean object itself, we can easily use java class reflection to call its method without having to reference to the client interface classes.(In WLS 6, reflection will not provide a way to use the bean without client-side interfaces.
Further more, in that every home interface for a stateless sessionbean is similar to this one containing only a simple create method, why not obviate it and just get remote interface by looking up JNDI?