- From: Noah Mendelsohn/Cambridge/IBM <noah_mendelsohn@us.ibm.com>
- Date: Wed, 19 Feb 2003 16:02:03 -0500
- To: "Martin Gudgin" <mgudgin@microsoft.com>
- Cc: "Adonis Amore" <infoletter@myrealbox.com>, xml-dist-app@w3.org, xml-dist-app-request@w3.org
Don't we need:
<xs:choice minOccurs="0" maxOccurs="unbounded">
<xs:element ref='tns:Fault' />
<xs:element ref='tns:Envelope' maxOccurs='unbounded' />
<xs:element ref='tns:Header' maxOccurs='unbounded' />
<xs:element ref='tns:Body' maxOccurs='unbounded' />
<xs:any namespace='##other' minOccurs='0' maxOccurs='unbounded' />
</xs:choice>
In other words, don't we allow more than one child of the body?
Actually, I have a deeper concern. While I would generally discourage
anyone from misusing our namespace, I'm not convinced this is the place to
enforce the restriction. In other words. I think I should be able to
send:
<soap:Body>
<soap:NoahsBogusSoapElement>
...
</soap:NoahsBogusSoapElement>
</soap:Body>
Furthermore, we don't even rule out sending a Fault with other elements,
we just indicate that it won't be recognized as a fault. No doubt, we
could change that, but it's the compromise we adopted and I don't want to
go back through CR for something like this.
All things considered, I think our original schema is the best compromise.
I believe it matches the prose in the case of both Fault and non-Fault
boby child elements.
------------------------------------------------------------------
Noah Mendelsohn Voice: 1-617-693-4036
IBM Corporation Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142
------------------------------------------------------------------
Received on Wednesday, 19 February 2003 16:37:06 UTC