Two of our RMI-IIOP API tests have started failing since rip-int b29:

1). Util.isLocal(Stub) should return true if the stub is in the same ORB instance as is the servant. However it is returning false.
  
2). Util.isLocal(Stub) should throw a RemoteException if the stub is not
connected to the ORB. It does not seem to be doing so.
 
Both these tests are reported as PASS for rip-int b28 and previous builds. It seems something has changed between b28 & b29.

As per bug-id 4395696, isLocal is changed to always return false. This is in clear conflict with 1.3 Javadocs for javax.rmi.CORBA.Util.isLocal which states

"Returns true if the stub is in the same ORB instance as is the servant".

Javadoc also mention about throwing RemoteException if the stub is not connected.


harold.carr@Eng 2000-12-21

Both of these symptoms are a result of the change I made to
Util.isLocal. In our merlin implementation it now unconditionally
returns false. This would cause both of the symptoms you noted.

So now the question is, is making this always return false compliant?
ptc/00-01-06 says

The isLocal method has the same semantics as the ObjectImpl._is_local
method, except that it can throw a RemoteException.

ptc/01-08 says:


The _is_local() method is provided so that stubs may determine if a
particular
object is implemented by a local servant and hence local invocation
API’s may be
used. The _is_local() method returns true if the servant incarnating the
object is
located in the same process as the stub and they both share the same ORB
instance.

The _is_local() method returns false otherwise. The default behavior of
_is_local() is to return false.

--------

It does not say the magic "shall" word so my take on this is that always
returning false is compliant.