- From: Amelia A Lewis <alewis@tibco.com>
- Date: Tue, 23 Nov 2004 15:34:06 -0500
- To: David Booth <dbooth@w3.org>
- Cc: www-ws-desc@w3.org
On Tue, 23 Nov 2004 14:04:37 -0500 David Booth <dbooth@w3.org> wrote: > The bottom line is that I suggest -- actually JMarsh made this > suggestion on the call, but I didn't manage to minute it in the midst > of our debate :) -- that the service be permitted to characterize the > fault either as a violation of its policies about where replies are > permitted to be redirected or as an MEP violation. How about letting > the service characterize the fault in whatever way it sees fit? Violent agreement. Sorry, I'm afraid that in the discussion, it must have appeared that Roberto and I were saying that the service MUST do something, specifically determine the node identity associated with both origination address and reply-to address. No. I think it *is* possible that a node could do so, and that, doing so, it could then feasibly fault with the reason "MEP violation." It could also have a set of policies, associated with or independent of node identity association with addresses, which could cause a fault in the same circumstances, certainly (as well as, potentially, in other circumstances; policy covers a wide territory). Both faults are possible. If our disagreement during the call was based on the notion that we would somehow require the service to perform some form of node-identity checking, then I must have misspoken. I would like the service to be *permitted* to fault in this manner, if, by means unspecified, it determines that the provided reply-to address is in fact *not* associated with the requesting node. That's all. That in turn suggests that best practice is to characterize exchanges in which the response is *expected* by both sides in the exchange to return to the requesting node (for some definition of node identity), but that if the service permits or expects the response to be directed to some third node, then a different MEP should be advertised. I think that most interactions are likely to be those in which the service expects the response to return to the requester, so that our publication of only that MEP is perfectly reasonable (although we *could* provide the additional third-party in-out MEP; it wouldn't be that difficult to show the binding in a non-normative note, for HTTP, using WSA or WSMD or both). Amy! -- Amelia A. Lewis Senior Architect TIBCO/Extensibility, Inc. alewis@tibco.com
Received on Tuesday, 23 November 2004 20:34:28 UTC