The rules are (from memory):
- Bean has a
@LocalBean
annotation -> bean has a no-interface view - Bean has a
@Local
annotation -> bean has a local view - Bean has a
@Remote
annotation -> bean has a remote view - Bean has no view annotations, but directly implements an interface which has a @Local annotation -> bean has a local view
- Bean has no view annotations, but directly implements an interface which has a @Remote annotation -> bean has a remote view
- Bean has no view annotations, but directly implements an interface which has no view annotations -> bean has a local view
- Bean has no view annotations, and implements no interfaces -> bean has a no-interface view
So, using @LocalBean
and using no annotation at all are both ways of getting a no-interface view. If you just want a no-interface view, then the simplest thing is not to annotate. Provided you’re not also implementing any interfaces.
Part of the reason @LocalBean
exists to add a no-interface view to a bean which also has an interface view. I imagine the scenario uppermost in the spec authors’ minds was one where you have a bean like:
@Stateless
public class UserPreferences {
public String getPreference(String preferenceName);
public Map<String, String> getPreferences();
}
Where you would want to expose both methods locally, but only the coarser-grained getPreferences()
remotely. You can do that by declaring a remote interface with just that method, then just slapping @LocalBean
on the bean class. Without it, you’d have to write a pointless local interface just to expose both methods locally.
Or, to look at it another way, the @LocalBean
exists because there is such a thing as a no-interface view, and the no-annotation option exists as a handy shortcut.