A few final potential issues

[Peter Niblett <peter_niblett@uk.ibm.com>]

While verifying some of ourstanding BaseN and BrokeredN issues, I have come
across a few more problems


WS-BaseNotification:

1. The schema type for the Notify Message currently looks like this

<xsd:element name="Notify" >
    <xsd:complexType>
      <xsd:sequence>
        <xsd:element ref="wsnt:NotificationMessage"
                     minOccurs="1" maxOccurs="unbounded" />
        <xsd:any namespace="##any" processContents="lax"
                 minOccurs="0" maxOccurs="unbounded"/>
      </xsd:sequence>
    </xsd:complexType>
  </xsd:element>

It looks like this fails the Unique Particle Attribution schema validation
test, since the wildcard can include a NotificationMessage. We should
change it to be namespace="##other"

2. The resolution of WSN 2.23 was that a Subscription does not necessarily
have to be a WS-Resource. This is indicated in words at the beginning of
Section 6. However there are still a few places in the text where we seem
to assume that all Subscriptions are WS-Resources. A quick search for the
string WS-Resource reveals examples at lines 301, 450, 650 (in the 02a
version).

3. The xsd for the SubscribeResponse element shows the
SubscriptionReference as optional, yet the pseudo-schema at line 640 shows
it as mandatory.. one or other of these is wrong. Are there any reasons why
an NP would want NOT to return a SubscriptionReference?

4. The wsdl includes a declaration for the xsi: namespace. This seems to be
redundant.

WS-BrokeredNotification:

5. The xsd/wsdl correctly refer to the element defining the
RegisterPublisher request wsn-br:RegisterPublisher, but the text in 6.1
shows it as wsn-br:RegisterPublisherRequest (in several places).

6. The RegisterPublisherResponse shows the PublisherRegistrationReference
as being optional (the pseudo-schema and real schema are consistent on
this). Depending on our decision on 3, we might need to change this.

7. The pseudo schema for PublisherRegistrationReference looks like this


       <wsn-br:PublisherRegistrationReference>
          <wsa:Address>
             Address of PublisherRegistration Manager
          </wsa:Address>
          <wsa:ReferenceParameters>
              PublisherRegistration Identifier
          </wsa:ReferenceParameters>
          …
       </wsn-br:PublisherRegistrationReference>?

    This (and the text that precedes it) appears to indicate that the
    <wsa:ReferenceParameters> is mandatory. We used to have something
    similar in the SubscriptionResponse, but we removed it to allow an NP
    to decide whether it wished to insert ReferenceParameters or not.

8. RegisterPublisher takes an InitialTerminationTime parameter, yet doesn't
throw the UnacceptableInitialTerminationTimeFault fault. This seems
inconsistent.

9. The wsdl includes a declaration for the xsi: namespace. This seems to be
redundant.


    Peter Niblett