Subject: RE: CPA composition from multi-role CPPs
Duane, the attached document looks very much like the TA document (lines 259 to 295 including TA figure 1). Few differences, mainly in referencing CPPs and CPA and in the bottom part of your doc. Here are my comments: 1. Beginning of second paragraph. When CompanyA "submits its own implementation details, reference links, and Business Profile information to the ebXML Registry" I think that you mean that it submits a CPP. Does it? 2. "business processes" (end of 2nd sentence of 2nd paragraph) and "business scenarios" beginning of 3rd sentence of 2nd paragraph) are they the same thing? 3. "usage of business scenarios" (end of 2nd paragraph). I concur with the comments from Marty. 4. I concur with Marty's comment on data transformation in paragraph 3 (this paragraph is not present in the TA document). In my opinion, in reality both parties would need some form of transformation but, likely, this transformation would be between the XML-form of some payload and the concrete internal representation used by a real-life legacy. 5. Paragraph 5. I commented in the real TA document the fact that it is not a requirement that the ebXML runtime software would be able to engage in any kind of automatic negotiation (including the simple exchange of CPPs). If here you are proposing a possible implementation, in addition to the bootstrapping issues highlighted by Marty, I do not honestly think that the state-of-the-art would make it possible to arrive to a comlete negotiation of a CPA performed by runtime software. In my understanding, the negotiation would be an off-line activity, which may be helped by some software but that would require at least a blessing from some human. 6. Paragraph 6 (fault tolerance). I think, as mentioned before, that the whole negotiation is an iterative process involving humans (helped by some smart "editor"-like application) instead than something that could be fully automated. So, the reversing of the roles would be part of the human-negotiation process and I would not go drilling into it to formalize an algorithm. But this is just my IMHO Thanks a lot for the attention /Stefano » -----Original Message----- » From: Duane Nickull [mailto:duane@xmlglobal.com] » Sent: 19 January 2001 21:22 » To: Martin W Sachs » Cc: Moberg, Dale; 'christopher ferris'; ebxml-tp@lists.ebxml.org » Subject: Re: CPA composition from multi-role CPPs » » » Here is the copy marked "for public distribution" » » Duane »
Powered by
eList eXpress LLC