Subject: RE: PartyId and Context
Henry I agree we have been through this loop once before in TRP. What is now happening is that other groups such as RegRep and TP are getting involved which is actually very good. I think that we should be able to come to resolution in a way which meets everyones needs soon. This will be good as there will then consistency across groups. Regards David -----Original Message----- From: Henry Lowe [mailto:hlowe@omg.org] Sent: Saturday, December 16, 2000 6:30 AM To: Burdett, David Cc: 'Duane Nickull'; Christopher Ferris; Scott Hinkelman; Charlie Fineman; ebxml-tp@lists.ebxml.org; ebxml-transport@lists.ebxml.org Subject: RE: PartyId and Context David, Agree 100%. But, as others have said, too, I thought we'd sorted this one out some while back. In fact, I was a bit of a stick in the mud at the time as I wanted to ensure that a URI would handle a CORBA IOR (it will). Best regards, Henry ------------------------------ At 01:45 PM 12/15/2000 -0800, Burdett, David wrote: >Folks > >I do hope this doesn't become as long a thread as last time, but here goes. >What I want to do is show a number of examples that describe what I think we >should do and then ask for views on whether this makes sense. > >EXAMPLE 1 >Consider this example: > ><From context=DUNS>45637284</From> > >What this means is that the number "45637284" belongs to the set of numbers >allocated by "David's Unique Numbering System". The problem I have is how do >you distinguish this from an **identical** From element where "DUNS" means >Dun & Bradstreets Numbering system? You can only do this if you now the >intention of the sender. Relying on the CPAId doesn't work since unless that >is also globally unique, then the number will be have been allocated by the >sender and you don't know who the sender is without looking at the CPA - a >circular argument. > >EXAMPLE 2 >Consider this example: > ><From>urn:duns.com:id:45637284</From> > >What this means is that someone has registered the domain name "duns.com" as >well as a urn structure to go with . If we go to IANA, we can see who has >registered "duns.com" probably Dun & Bradsteet we then know **completely >unambiguously** who allocated the number and what it means and who it >identifies. "David" would not be able to validly use "duns.com" since it >would have already been "taken" by Dun & Bradstreet. > >EXAMPLE 3 >Consider this example: > ><From usercontext=DUNS>45637284</From> > >If we define usercontext as meaning "a code that identifies a set of numbers >that have allocated or devised by a party" ... with the following >explanation ... "the values of usercontext are not necessarily unique. Two >or more parties may allocate the same value and associate it with different >sets of numbers. This means that sender or recipient of a message that >contains a usercontext, MUST be sure that a recipient of the message knows >unambiguously the party that is being identified. How this is done is >outside the scope of this spec". > >SUMMARY >What I propose is that we allow both example 2 and 3. Specifically: >1. If no usercontext is present then the content of the From (or To) must be >a URI. >2. If usercontext is present then the recipients of the message must be able >to unambiguously determine who sent the message by methods mutually agreed. > >Does this make sense? > >David > >-----Original Message----- >From: Duane Nickull [mailto:duane@xmlglobal.com] >Sent: Friday, December 15, 2000 12:23 PM >To: Christopher Ferris >Cc: Scott Hinkelman; Charlie Fineman; Burdett, David; >ebxml-tp@lists.ebxml.org; ebxml-transport@lists.ebxml.org >Subject: Re: PartyId and Context > > >All: > >I see problems arising with the use of one and one only method for >generating a Party ID. We need to allow multiple schemes for >identifying parties. > >In the cases where multiple SME's are using a single ASP for a web based >onramp to the ebXML infrastructure, but they may wish to change ASP's or >eventually get their own system, this will not work. When a company >looses their URI - do they loose their Party ID? > >URI's by themselves will not work. They could be one such method for >uniquely identifying a party but not the end all and be all. > >Duane Nickull
Powered by
eList eXpress LLC