I didn't allow reconnect either although I could have done it easily since I had a ServerEventService (which was also bound in the RMI registry) that ran in a background
thread. There were four types of ServerEvents: startup, delayedShutdown, shutdown and heartbeat. A heartbeat was sent every 60 seconds for client confidence that the server was alive. If five heartbeats were missed, then the client assumed the server was dead. The problem with a true network failure is that in order to reconnect, you would have to "poll" the service periodically ignoring RemoteExceptions until the service came on line again. I just felt like it was better to inform the user of the stale connection and suggest that they try connecting later.
Hope this helps,
Michael Morris
Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius - and a lot of courage - to move in the opposite direction. - Ernst F. Schumacher