- From: Philippe Le Hegaret <plh@w3.org>
- Date: 10 Apr 2003 08:42:00 -0400
- To: www-ws-desc@w3.org
abstract level: The goal of defining them at the abstract level as well would be to provide a _hint_ to the implementation. Their use is not required. Mapping XML Schema complexType into an HTTP GET won't be done. Of course, you can serialize the Infoset in the URI but who would like to use those uris? Based on the abstract feature, one can determine if a result is cachable for example, even if an HTTP POST is used underneath. feature name http://www.example.org/CRUD property name: http://www.example.org/method value space: create | retrieve | update | delete binding level: HTTP binding: feature http://www.example.org/2003/03/http/web-method property name: http://www.example.org/2003/03/http/web-method/method value space: PUT | GET | POST | DELETE SOAP HTTP binding: feature http://www.w3.org/2002/12/soap/features/web-method/ property name: http://www.w3.org/2002/12/soap/features/web-method/Method value space: PUT | GET | POST | DELETE issue: Do we need to invent a new uri for for the HTTP binding feature? Well, not really, but it would be good if people don't associate systematically the SOAP Web Method Feature with SOAP itself. They may do since the uri provided in the SOAP specification contains "soap". It's more a feature that needs to be attached to HTTP itself but the URI used does not reflect that enough imho. issue: and what about the HTTP headers? How useful would it be? The accept, accept-ranges, content-type, authorization, cache-control, connection and content-length are already fixed by other means (security, authorization features). accept-language has nothing to do in the WSDL. Do we have an example of an header that needs to be fixed and should not be represented in a more abstract way? Philippe
Received on Thursday, 10 April 2003 08:42:01 UTC