RE: On WSDL "operation"


> If so, then the only application semantics you would need are;
> - "submit document", where success means "submitted, thanks", and
> - "request document"

Services only know about messages. When you send a message to a service,
you don't know whether there is a cached response to your message,
whether a human who is a fast typist creates that response, whether a
program creates "garbage" that just adheres to the structure of the
reply, or whether an operation or a process is executed. I don't think
that by saying that services are defined in terms of the messages being
exchanged suggests any application semantics, even the simple ones that
you suggest. In fact, that was one of the points... application
semantics should not be suggested. It is not the purpose of WSDL to
suggest such semantics. If one thing the WS community does well is to
come up with specifications for almost everything. So, there are
specifications out there that would allow service architects to define
semantics (DAML-S perhaps?). (BTW, my favourite latest WS specification
proposal is the "Framework for WS Implementation" for which a TC has
been formed at a standards organisation near you... where is this world
going to? :-)

> That's HTTP POST and GET, respectively.

That's one of the possible message exchanges. The fact that this
particular message exchange is captured by the more general case should
be an indication to the appropriateness of the term "messageExchange"
rather than "operation". The latter does not capture the essence of the
message exchange you suggest.

> So I agree with you, but it doesn't appear that you appreciate the
> implications of what you're suggesting.

Your message did not suggest agreement. However, I am curious to the
reason you think that Jim does not "appreciate the implications" of his
suggestion. If there is a need for clarification in what we are
supporting, please let us know and we'll do our best to explain it
(although I think that we and others have covered this issue). If the
community thinks that we are wrong (and I have definitely been wrong in
past discussions) we will just accept it and move on to the next topic
of discussion (perhaps a more important one:-)

BTW, I liked Amy's proposal ("messageExchangeInvolvingAService") but
perhaps we could make it even more specific...
"messageExchangeWithoutAttachedSemanticsInvolvingAService" :-)


Received on Wednesday, 17 September 2003 11:03:00 UTC