ebxml-msg — archive

RE: [ebxml-msg] Message delivery to non-adressable endpoints.

From: Jacques Durand
To: [email protected]
Date: Wed, 17 May 2006 19:51:00 -0000

MHonArc v2.5.0b2 -->

ebxml-msg message

Subject: RE: [ebxml-msg] Message delivery to non-adressable endpoints.

From: "Durand, Jacques R." <[email protected]>
To: "Dale Moberg" <[email protected]>, <[email protected]>
Date: Wed, 17 May 2006 12:51:09 -0700

Dale:

Some polling capability is considered in WS-RX but no agreement yet between proposals. Some proposal is only within the context of pulling for reliability purpose. Another is more general but then might as well (should?) be a separate specification, as argued now in WS-RX.

WS-add plays more or less of a role in these proposals.

ebMS PullRequest is a very simple signal that has not much additional semantics than a general polling signal – and sure could be replaced. Now the question is how this kind of reuse plays in a context with several SOAP nodes serialized on an endpoint, each possibly holding some messages to be pulled (which is the case with ebMS, and its RM module).

The no-reuse solution means a distinct signal is targeted to each node (e.g. pulling RM-specific messages separately from ebMS messages).

An all-reuse solution would use a single “generic” poll signal – but we need to see how it will be controlled or targeted separately for each spec / each node that is on its way (can the same signal be targeted to more than 1 node so that the same underlying response will piggy-back all responses, does it only need to distinguish these, on the initiator side, which SOAP node will initiate the polling? All? Just one?)

So a reused-poll-signal solution sounds attractive but even if one is available today (open standard track, no IP strings attached…) we need to define first how that is expected to work in a multi-node env. It’s a bit of un-chartered territory here.

Meanwhile a step in a reuse direction could be to not exclude the use of other poll signals, that could be later “profiled” for ebMS. Of course either a conformance profile or a P-mode needs to nail down the interop agreement on this.

Jacques

From: Dale Moberg [mailto:[email protected]]
Sent: Wednesday, May 17, 2006 9:19 AM
To: [email protected]
Subject: [ebxml-msg] Message delivery to non-adressable endpoints.

I am trying to catch up on reading WS-* groups that we are utilizing in ebMS 3.0.

The WS-RX group is discussing how to make use of wsa to do something that, at least to me, resembles poll/pull MEP.

Since one of our goals is to converge ebMS with new WS-* standards-body-approved specifications for functionality fulfilling ebMS requirements, I am wondering whether we should consider reusing wsa to do this rather than make our own signal. The MPF identifier can be probably squeezed into a field for metadata extensions.

Before I contribute a public review remark on this, does anyone on the TC have some comments on this?

Thanks

Dale Moberg