W3C home > Mailing lists > Public > www-ws-desc@w3.org > September 2002

Re: Rationale for Dropping the <soap:body use=...> Attribute

From: <ryman@ca.ibm.com>
Date: Wed, 18 Sep 2002 09:59:26 -0400
To: Jacek Kopecky <jacek@systinet.com>
Cc: WS Description WG <www-ws-desc@w3.org>
Message-ID: <OFC2A5A470.38E2831B-ON85256C38.004BDE3C@torolab.ibm.com>


I agree that the original authors of the WSDL had a lot of generality in
mind concerning multiple type systems. However, I prefer the viewpoint that
messages should be described abstractly, so all you really need is one
sufficiently expressive type system. XML Schema fills that role. While it
is best for tree-like data, it can also be used for graphs via the ID and
IDREF types.

All details about how the message is formatted should be in the binding.
For the SOAP binding, we are proposing that the only binding we need is
literal. Attempts to use a more flexible bindings, SOAP encoding, led to
interop problems.

So, no, I'd say not to use a different kind of schema language. Leave the
message definition independent of the binding. If more flexibility is
really needed, then modify the SOAP binding rules, but specify the encoding
algorithm more clearly to eliminate interop problems.

Arthur Ryman

                      Jacek Kopecky                                                                                                                 
                      <jacek@systinet.c        To:       Arthur Ryman/Toronto/IBM@IBMCA                                                             
                      om>                      cc:       WS Description WG <www-ws-desc@w3.org>                                                     
                                               Subject:  Re: Rationale for Dropping the <soap:body use=...> Attribute                               
                      09/18/2002 07:39                                                                                                              

 Arthur, others,
 I agree that use="literal" should be sufficient in WSDL. My
understanding has always been that for other datamodel than the XML Tree
datamodel of XML Schema, other schema languages should be used in WSDL
descriptions - something like

<soapenc:schema targetNamespace="foo" xmlns="foo" xmlns:foo="foo">
  <soapenc:struct name="myStruct">
     <soapenc:edge name="a" targetType="myStruct"/>
     <soapenc:edge name="b" targetType="myArray"/>
     <soapenc:edge name="c" targetType="soapenc:simple"/>
  <soapenc:array name="myArray" dimensions="3">
     <soapenc:edge targetType="soapenc:simple"/>

and then a message part of type foo:myStruct would be literally
serialized according to its schema, whichever variant of the actual XML
instance is the result.
 Is this how others see the issue, too? 8-)
 Best regards,

                   Jacek Kopecky

                   Senior Architect, Systinet Corporation
Received on Wednesday, 18 September 2002 09:59:32 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:54:39 UTC