- From: Savas Parastatidis <Savas.Parastatidis@newcastle.ac.uk>
- Date: Fri, 20 Jun 2003 22:48:29 +0100
- To: <www-ws-desc@w3.org>
David, [snip long, very helpful message] Thank you very much for the example. It was enlightening. So, I guess it's a matter of preference. The @targetResource introduces identity to a web service resulting in a great number of services, as in your example. You end up having 3000 printer services and 5000 management services. Personally, I see Web services as coarse software agents. I see only one printer service and only one management service. If you want to locate the printer that is closer to you, you go to a printer registry (perhaps through the management service or a printer registry service) and you get back a printer id, an XML instance that adheres to the Printer XML Schema defined by my organisation. The Printer XML Schema I got back can be passed as an argument to the operations supported by my printer and management services. Your RDF and registry examples are still valid. Instead of finding services, though, you find printer tokens, ids, URIs (call them as you want). Such an approach is closer to the document-exchange model of Web services. If my understanding is correct, the targetResource encourages services to be seen as objects rather than coarse software components. We move to an O-O based approach instead of a component based one. I can see the value in both approaches. Regards, .savas.
Received on Friday, 20 June 2003 17:48:32 UTC