When should remote managers reset their counterparts?

[GoogleCodeExporter]

I need to determine when a server-side remote manager should try to reset 
its counterpart.

For example, can the reset() safely forward an init() to its remote 
counterpart?  

What happens when the counterpart is also the self?

Are there managers that would prefer not to reset automatically?

How dirty does a remote counterpart get?  I can imagine that with multiple or 
extended runs, a remote counterpart could accumulate a lot of stale objects, 
which might lead to poor map access times.  Are there other problems?

Original issue reported on code.google.com by Benjamin.Heasly on 16 Feb 2010 at 10:48

[GoogleCodeExporter]

This should come up when we get into serious usage and in-lab testing.

Original comment by Benjamin.Heasly on 16 Feb 2010 at 11:12

• Added labels: Priority-Low

[GoogleCodeExporter]

As of r237, client managers request a clearObjects from the serverManager, but 
dots not attempt a full initialize.

Original comment by Benjamin.Heasly on 23 Jul 2010 at 11:41

• Changed state: Fixed