- From: Savas Parastatidis <Savas.Parastatidis@newcastle.ac.uk>
- Date: Mon, 15 Sep 2003 10:18:31 +0100
- To: "David Orchard" <dorchard@bea.com>
- Cc: <www-ws-arch@w3.org>
David, Coming up with a definition for SOA and the differentiating aspects (if any) from Object Orientation is, indeed, hard. Unlike you however, I believe that there is a difference... Fundamental to O-O is the concept of type sharing. With distributed objects, it is necessary to share type information throughout the distributed application. I see SOA as a paradigm where type sharing is not required. XML and XML Schema give us the common understanding to describe messages. We need to agree on message exchange patterns rather than on interfaces. One may argue that the well-defined structure of messages is equivalent to having well defined types. I wouldn't agree with that but, then, that's just me. When you have types you start talking about references or pointers to instances of those types (as you hint in your message). That leads to tighter coupling than the SOA paradigm suggests. [snip] > > In the past, people have typically differentiated SOAs from d-o based upon > best practices, such as "soas have a well published address" - this > differentiates from d-o factory pattners, and "soa interactions are coarse > grained" - this differentiates from the "setter/getter" style of d-o. But > these are simply best practices and not constraints. Another way to > differentiate is the format. > [snip] So, I don't see the difference between SOAs and D-Os as just of a "best practise" one. I believe that SOA suggests a paradigm that may not be a revolution over distributed object-orientation but I support that it is definitely an evolution; an evolution that did not come overnight (e.g., CORBA services). Regards, .savas.
Received on Monday, 15 September 2003 05:18:36 UTC