This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
EVD defines an eventType as: <wsevd:eventType name="xs:NCName" element="xs:QName" actionURI="xs:anyURI"? ...> xs:any* </wsevd:eventType> Notice that the "element" attribute is required. Normally this makes sense since we would expect at least some data to appear in the soap body. However, it seems like it might be possible to have situations where the only thing that's needed is the actionURI on the notification msg. For example, wsman defines hearbeat messages that an event source might send periodically so that the sink knows the subscription is still alive. In these cases the soap Body isn't needed at all so does it make sense to require someone to define a GED in these cases? Proposal: Make "element" optional. Add "OPTIONAL" to the defn of @element: /wsevd:EventDescriptions/wsevd:eventType/@element This _OPTIONAL_ attribute refers to a GED defined or imported in the /wsevd:EventDescriptions/wsevd:types element. The referenced GED serves as the definition of this Event Type. and modify the xsd to match. Also, in eventing we would need to tweak the EVD binding sections. for example (see *** text): The [Body] property of the notification has a single child element. This child element is an instance of the Global Element Declaration referenced by the element attribute of the wse:eventType element corresponding to the type of the event being transmitted. *** If the element attribute is absent then the [Body] property has no children. *** and The /soap:Envelope/soap:Body/wse:Notify element has a single child element. This child element is an instance of the Global Element Declaration referenced by the element attribute of the eventType element corresponding to the type of the event being transmitted. *** If the element attribute is absent then the wse:Notify element has no children. ***
My concern is this: This means that implementations MUST understand the action URI's at the application level, just incase the body is empty. I have no real objection, but we should make sure of the ramifications. This might also imply some "MUSTs" on the receiver to deal with the options.
Dave - I think the concern you have is actually more related to the definition of formats since someone could define a format that doesn't use the @element attribute at all. No?
Thinks about adding text to say that a format URI needs to talk about what to do about missing fields - like actionURI
2010-04-20 resolved as proposed with the amendment to add: at least one of the two (action / element) MUST be present