|
Uh, that bullet should start "A Notification _Consumer_
may...", no?
-- ReC
Attached please find a revision of the current committee draft, as
per my action item from the last conference call. Change tracking is
turned on, so it should be easy to see what I've done. There are two main
changes, and a few more following from those (and perhaps a few more such
changes that I've missed):
- I've added this bullet item to the definition of
NotificationConsumer in section 2: "A Notification Producer may be a physical
web service, capable of receiving messages over a network, or may be a
virtual endpoint known only to a given set of
NotificationProducers. For example, the “Pull Points”
described in section 5 may be virtual. In either case, the
NotificationProducer will use the NotificationConsumer, together with other
information in a subscription request, to effect the desired handling of
Notifications, or will fault if this is not possible."
- I've changed the first sentence of section 5.1.1
(accumulating Notification messages) from "The PullPoint interface supports
the NotificationConsumer interface (as defined in section 3)."
to "A PullPoint produced by the CreatePullPoint method MAY support the
NotificationConsumer interface (as defined in section 3)."
This may or may not imply changes to the WSDL, and we should discuss this.
My natural inclination is to be a bit nervous of introducing a
"physical/virtual" distinction here, but I believe that such a distinction
is already implicit, and introducing it does not require massive changes to
either the semantics or the text describing the semantics. Indeed, one
could argue that it simply changes the text to make the implications of the
existing semantics more explicit.
In any case, the proposed text is very
much open to further discussion, and may well prove not to be the best way
forward.
|