[Modeling] Modeling an Agent Class - alternate proposal

[James Odell]

On 6/24/03 5:49 PM, "Wagner, G.R." indited:

> Why not first go for a less drastic extension: split up Class
> into NonagentiveObjectType and AgentType (so, everything is
> still an object:-)?

Either way, we still need to define what goes into an AgentType.  We have to
do that before we can define the NonagentiveObjectType; i.e., I don't know
how to subtract nonagentive qualities is I don't know what agentive
qualities are.  So, if we focus first on defining what an agent type
metaclass is by starting as a subclass of Classifier, we can make important
progress.  Then, after it's defined, we can do some refactoring.  Gerd, if
you already have a proposal as to how the NonagentiveObjectType and
AgentType might be defined, then please send it in.  We need all the help we
can get.

-Jim