Re: Encapsulating Derivation

The FE reports, in the FE Object LFB, each LFB class it supports.
The only way a CE "queries" something is by getting some LFB information.

The obvious way to support this is:
The Supported LFB table can indicate the parent class.  (We would 
probably need a virtual flag that indicates that the FE will not 
instantiate directly certain parent classes.)  There are several 
possible representations for the table.  (A tree, a parent class 
listed with each class, duplicating a little information but making a 
simpler list, ...)

I considered trying to add this information as properties of the 
LFB.  But to do a GET-PROPERTY there has to be an instance.  And the 
CE would need this information before it created an instance.

If there is some other good place to put this information, I 
certainly am open to alternatives.

Yours,
Joel M. Halpern

At 11:27 PM 10/18/2006, Jia Fenggen wrote:
>I agree that we should substitutability and i wonder should we allow 
>the CE to query the parent class list when it encounter a LFB that 
>it don't know?Has their any method in protocol draft to do these,thanks.
>Yours,Fenggen