ubl-ndrsc — archive

[Kris Ketels]

maybe a stupid question, but what is the relationship (if any) between this discussion and the 'generic header' inititiative going on in the UN/CEFACT ATG2 group? "Burcham, Bill" wrote: > Thanks for the clarification Eduardo. If that's the case then I don't see > the big win here. The old ebXML-ers (:-) were right when they divided the > world into messaging, document, process, agreement, regrep. Separation of > concerns is a Good Thing. Encapsulation is our friend. And XML, with its > simple, regular, universal structure (the well-formed document) is perfectly > capable of supporting the combination of UBL documents as currently > architected, with whatever "generic header" structure or generic envelope > structure someone creates in the future. > > I think the notion that the generic header design must infect er affect the > document design is an artifact of pre-ebXML EDI-thought. If you look at X12 > with its mixture of transport, transaction, and document structures its > pretty easy to see that yes, if you want to inject "generic messaging > headers" in there you're gonna have to make space above the document level. > Which means you're gonna have to change the X12 standard. Same for EDIFACT. > > UBL on the other hand has no transport or transaction level information. In > the spirit of ebXML, UBL defines only document level terms. Therefore, the > only logical place to inject (messaging) headers would be outside the > top-level UBL elements. That being the case, UBL can remain blissfully > decoupled from the generic header effort while fully supporting it. No > action is required on the part of the UBL designers. > > If your messaging system needs to pass on generic header information to a > back-end system, it can simply pass a document that looks like this: > > <foo> > <generic-header-container> > ... > </generic-header-container> > <ubl:Order xmlns:ubl="..."> > ... > </ubl:Order> > </foo> > > -Bill > >