OASIS ebXML Messaging Services TC

 View Only

RE: [ebxml-msg] Have you implemented Version 2.0 ?

  • 1.  RE: [ebxml-msg] Have you implemented Version 2.0 ?

    Posted 03-08-2002 16:08
    Title: RE: [ebxml-msg] Have you implemented Version 2.0 ? Ralph, thanks for the kind remarks.  You are correct that our main objective was Interoperability rather than Conformance, but this does not mean we ignore conformance.  While it is probably not possible to assure 100% conformance to any specification, I would venture the opinion that the DGI method actually gives greater assurance of conformance than any other.  We certainly scrutinize the output of every participant's code to make sure there is no deviation from the specification.  Since we had four participants this round, this meant that there were at least four sets of eyes checking each company's submission for any such deviation.  The more eyes the better.   The only other method I know of where conformance may be tested is some form of Reference Implementation.  This was the method used for years, primarily because it was not feasible to do the kind of rigorous, full-matrix testing now possible on the Internet.  After all, a Reference Implementation is still an Implementation and subject to flaws.  Even though, in the past, this was the accepted way to assure conformance (because there was nothing better), this method does not always to produce satisfactory results.  Even if product A works with Ref and product B works with Ref, there is no assurance that A works with B -- and it is relatively sure they don't.  DGI estimates that you will get 85-90% Interoperability with this method, which is a good start but not nearly enough for B2B.   In terms of cost, Reference Implementations for this type of application typically cost in the millions, plus maintenance.  DGI is providing a comparable (superior) method for less than a tenth of that cost.   You are correct, we do not focus on Compliance, but that is included in our end result.  Please ask anyone in the test whether we are out of compliance with any part of the spec?  Because of time and resource constraints (and a wandering specification), we were forced to test a subset of the total spec in this round, but that will be remedied in the future.   Regards,   David Fischer Drummond Group.