[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: [cam] FWD: RE: Use of pre- and postconditions / inconsistencies inspec?
Just an FYI here - on some interesting interchange today on the CEFACT BPSS side. I had no idea they had backed themselves into a corner on this one! I will let every know if they decide to make an implementation note in V2.0 BPSS on this approach. DW. -------------Forwarded Message----------------- From: "UN/CEFACT TMG Business Processes WG", INTERNET:uncefact-tmg-bpwg@listman.disa.org To: "UN/CEFACT TMG Business Processes WG", INTERNET:uncefact-tmg-bpwg@listman.disa.org Date: 3/11/2003 3:22 PM RE: RE: Use of pre- and postconditions / inconsistencies in spec? JJ, I did not realize this was an issue - so good this came up! An alternative here is to push this logic down into the business transaction layer. So that the business transaction layer returns a "True / False" response, plus an optional error status - back up to the BPSS layer. This certainly allows the BPSS to continue to operate with that binary logic approach - and then the ebMS layer can carry the error response back to the sender. Also - then the overall process status stuff is carried at the application layer - on which its going to be highly dependent, and where I'd say it makes most sense - instead of the BPSS having to provide tracking mechanisms. Better yet - from the CPA / BPSS interface perspective - you need only note that the error response mechanism needs to be supported. I would also note that the OASIS CAM mechanism already fully specifies this behaviour - and can access context variables from the BPSS layer to further facilitate this (and those context values can get passed thru from the ebMS layer - on a URL if necessary). Also - by inspecting the validation section in the CAM template that your partner is using - you can see exactly the end condition checks and business rules. Part of the CPA agreement requires alignment on those transaction business rules anyway - so this IMHO makes an excellent fit. Check out the latest draft at: http://cam.swiki.net/14 and see the section on transaction validations and the context variables. Thanks, DW. ============================================================ Message text written by "UN/CEFACT TMG Business Processes WG" > One of the major issues is that there is no official syntax for such conditions. Another one is that they allow to specify unilateral specifications, a party could specify that a purchase needs to be approved before it can be sent out but that "state" is unreachable by the other party which has no way to know if the order was approved before it can decide whether it is valid to receive the purchase order. In the final 2.0 version, the plan (that needs to be discussed) was to limit the scope of pre/post conditions as annotations (e.g. if I send you a PO, it means that this PO was improved internally). In particular, it is forbidden to express choreography constraints based on them. The only way to specify choreography in 2.0 would then be transitions between BTAs and condition expressions on the transitions based on the content of the documents or the type of documents exchanged during the collaboration. <
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]