- From: Mark Baker <distobj@acm.org>
- Date: Thu, 23 Oct 2003 21:45:53 -0400
- To: Savas Parastatidis <Savas.Parastatidis@newcastle.ac.uk>
- Cc: www-ws-desc@w3.org, Jim Webber <jim.webber@arjuna.com>
Thanks for your response, Savas, it was very helpful. I'll just cut to what appears to be the source of my confusion ... On Fri, Oct 24, 2003 at 12:01:01AM +0100, Savas Parastatidis wrote: > I wouldn't call the above description as a "document endpoint" but > effectively we are talking about exchange of documents (the messages). Ok, but you're really talking about the exchange of two very different kinds of documents; ones that contain only state (e.g. "12345"), and ones that contain an operation and state (e.g. "add(12345)"). My concern and confusion stems from the two apparently being treated the same. As a concrete example relevant to interoperability, consider the semantics of a successful SOAP response to each of those two example messages above. For the first example, the response message can be interpreted to mean "that data is successfully processed", which could mean many things, but that's hidden from the client. But in the second example, what does success mean? Does it mean that the data was processed, or does it mean that the number 12345 was added to whatever the existing state was? It could mean both. And when you combine that with the issue Umit is talking about, it's even more ambiguous, as it could mean something else entirely because the operation may not be in the message. i.e. it could mean the same as "add(12345)", though only "12345" is on the wire. Yikes! Mark. -- Mark Baker. Ottawa, Ontario, CANADA. http://www.markbaker.ca
Received on Thursday, 23 October 2003 21:45:37 UTC