On Thu, Nov 19, 2009 at 4:00 PM, Peter Saint-Andre <[email protected]> wrote: > On 11/19/09 1:54 PM, Andy Skelton wrote: >> Next up, a request on behalf of publishers. If the publisher is not >> allowed to determine the version, shouldn't it at least be able to >> learn the resulting version when it publishes, modifies, or retracts >> an item? > > Yes. That would be in the notification, as above.
I saw that there but it didn't seem enough. The notification goes out to subscribers. Publishers are not necessarily subscribers. The relationships are orthogonal. The publish IQ (A) changes a state (S -> S') and then gets a response (A'). Say a publisher needs to know the state S' resulting from A. The natural way to return this new state is via A'. Subscribing to notifications (N) allows the publisher to learn about each new S but to say that A caused S' because N(S') followed A is a leap of inference and a heap of extra logic. Can we please just return the new version in the IQ response? :-) Andy p.s. I don't have a use case where the publisher needs to know its versions. I just think it's the right thing to do.
