RE: Web Service Definition [Was "Some Thoughts ..."]

I have been following this discussion for a while an I would
like to give my opinion. I anticipate that this is my personal
opinion and that does not necessarily reflect the one of my
employer.

1.	I think that it will be difficult to find a definition for 
	w/s without definining which properties a w/s should exhibit.

		As an example, in other branches of the science, 
		categories are defined (if I am not wrong) starting 
		from some taxonomies. So, one could, for instance, 
		spell the definition of a "mammal" starting from a 
		categorization which takes in account the properties 
		for that category.

	For this reason, I think that the effort started by Joseph
	Hui is the right one; once we have defined the properties a
	w/s should exhibit, we will be able to define a w/s as
	something possessing such properties.
	As well, the recent posting from Edwin Khodabakchian ("Putting
	W/S in context" is, imho, a good starting point.

2.	For a component to become a w/s, it is important, imho,
	that it exhibits a behavior that is predictable, clear
	and complete. We may have simple things that do not
	need the whole richness, but they could probably only
	be used in certain situations.

	Architecture should, imho, operate at a more behavioral level;
		- simple protocol interoperability risks not to 
		  capture what is important for runtime and behavioral 
		  interoperability of w/s
		- description of entry points may not always be
		  enough to define how a w/s operates in a given
		  context

3.	I also think that in many of the previous mails, there was
	a sort of "implicit assumption" that a w/s is something
	which "responds" to inputs. 
	Whilst this is an important category of w/s, I think that
	a w/s may also be capable of initiating conversations
	or carrying out conversations in an async way.

Best regards
/stefano
	

Received on Tuesday, 5 March 2002 04:19:03 UTC