- From: Martin Gudgin <mgudgin@microsoft.com>
- Date: Tue, 25 Jun 2002 19:18:47 -0700
- To: <www-ws-desc@w3.org>
I took an AI at the last telcon to write up Issue 25. Here is that write up. The issue is regarding the relationship ( or lack thereof ) between XML Schema and the SOAP Data Model. WSDL 1.1 allows rpc/encoded endpoints to be described using XML Schema type definitions. Unfortunately the 'encodedness' of the message means that the schema type often does not actually describe the XML that is placed on the wire. This is akin to saying 'You see that schema type over there? Well the XML doesn't look like that!' The problem stems from the fact the SOAP 1.1 data model ( not that there is such a beast formally ) describes a directed edge labeled graph while XML Schema descibes edge labelled trees. In SOAP 1.2 we addressed this issue by seperating XML Schema out from the SOAP Data Model and stating what would happen given certain scenarios; 1. Reading a message with no associated schema 2. Reading a message and validating with <xs:schema xmlns:xs='http://www.w3.org/2001/XMLSchema' /> 3. Reading a message and validating with http://www.w3.org/2000/12/soap-encoding.xsd 4. Reading a message and validating against a message specific schema At the level of the SOAP data model all that happens is that the values of the type properties of graph nodes get more precise as you move from scenario 1 to 4. I'm not sure what to propose for WSDL. Ideally we would clean up the encoded stuff so that it could be described accurately using XML Schema but I think that's probably a bigger job than we want to tackle right now. Ditching encoding would do the trick but I don't know what other people think about that. Discussion? Gudge
Received on Tuesday, 25 June 2002 22:19:18 UTC