RE: DAML-S expressiveness challenge #1

From: Mitch Kokar [mailto:kokar@coe.neu.edu]
> If DAML is to be used to describe programs (at a high level) 
> then it seems to make sense to use the representations that 
> the software engineers use for
> this purpose. This representation is UML.  UML has a number of ways of
> representing behavior: sequence diagrams, collaboration 
> diagrams, state
> transition diagrams and activity diagrams. I believe it is 
> better to adopt
> some or all of these representations rather than invent new ones.
I disagree. That would be like saying that companies today should advertise
in the yellow pages by posting the flow charts of their business processes.
Consumers choose vendors by *what* they do, not how they do it.  Similarly,
agents (e-services, web-services, etc ... pick your poison) will also want
to choose other services based on what they do, i.e. on the pre- and
post-conditions, and meta-data such as cost and quality. So a state
transition diagram would simply be the wrong level of detail.  In any case,
a graphical notation is not really suited to automated processing.

Cheers,
Ian

____________________________________________________________________
Ian Dickinson    HP Labs, Bristol, UK    mailto:Ian_Dickinson@hp.com