Re: The requirements on message 'correllation'

Andrew Berry wrote:
'the "original order" and "replacement order" should have 
distinguished names in the program, even if they transmit the same 
information, because they have significantly different semantics.  This 
unambiguously defines the semantics without constraining the binding to 
particular technologies.  Note that doing it at this level also makes 
it relatively easy for a compiler/verifier to detect and flag 
ambiguities.'

I fully and enthusiastically agree.  

"Correlation" has always seemed like a hack to make up for a lack of
semantics.
(I know, it's a useful hack, but still...)

Semantics of pre- and post-conditions are related.

So are in fluents in event calculus.

Assertions in the domain of discourse (e.g. business) make the states of
a choreography more meaningful to both technicians and end users, as
well as providing the benefits Andrew cites above.

Received on Wednesday, 10 September 2003 07:30:21 UTC