ietf-openproxy
[Top] [All Lists]

RE: OPES ARCHITECTURE for rule processing and service execution

2001-02-12 15:44:54
Actually I am not sure exactly how we should define conflict. Something
like:
        Rule #1) if A, do P.
        Rule #2) if B, not do P.
Then when both A and B are satisfied, we have a conflict. But if A and B are
not satisfied at the same time, then we don't have a conflict. So we can not
do much static analysis in this case. At run time, how do we resolve it?
Today's OPES drafts suggest using ordering to take care of it -- so we don't
really resolve the conflict per say.
Another example, a CP asks for "ad insertion", then local access provider
asks for "regional ad insertion" again, and then the client might ask for
"ad stripping" after all that is done. Do you consider this kind of semantic
contradiction a conflict? It would need some intelligence at the sematic
level, not just syntax level. 

Simply dictate an order between the rule modules and within the rule module
and let the ordering take care of it is a pretty simple solution to all
these. Is it sufficient?

anybody with good examples on the possible conflicts ? 

Lily

-----Original Message-----
From: Rajnish Pandey 
[mailto:Rajnish(_dot_)Pandey(_at_)india(_dot_)sun(_dot_)com]
Sent: Monday, February 12, 2001 12:45 PM
To: lily(_dot_)l(_dot_)yang(_at_)intel(_dot_)com
Cc: ietf-openproxy(_at_)imc(_dot_)org
Subject: OPES ARCHITECTURE for rule processing and service execution



Hi,


  ---------------------------------------------
  Issue No 4:
  
  How exactly do we resolve the conflict in Rule Base?
  
  Rule base should be designed in such a way that only the 
relevant part of the 
rule base is considered for rule matching .
  
  
 Rajnish