Carl,
Precisely.
And these authorities usually already have their own coding scheme
worked out already - so you really want to make sure you have
that "External ID" slot and mechanism to just put that straight in.
No need for us to create extra work for people -just leverage
what they have done already and automate the use of it.
Behold the power of UID.
Cheers, DW.
----- Original Message -----
From: "Carl Mattocks" <>
To: "David RR Webber" <>
Cc: <>; "Chiusano Joseph" <>;
"Duane Nickull" <>; <>;
<>
Sent: Friday, April 09, 2004 3:31 PM
Subject: Re: [regrep] Issue with UUID's for Core Components and BIE's
> Yes, Yes - similar to the way the Getty Art Foundation acts as the
> authority for the semantic relations embedded in Art & Architecture
> Thesuarus and ACORD is the authority for Schemas used in Insurance
> transactions (and eventually the ontology) ...
> I expect the content of many Registry / Repository services will be
> organized using 'associations' defined elsewhere and therefore require
> synchronization on a regular basis. .
>
>
> <quote who="David RR Webber">
> > Carl,
> >
> > You raise an interesting another point. In BCM we talk about
> > domains, communities of interest and then authoritive sources.
> >
> > Such an authority clearly has to be responsible for managing the
> > UID values within their domain.
> >
> > Using my example - the USPS is managing the semantics
> > around their address formats. Then this can be crosswalked
> > to the international authority - the UPU - and their UID set,
> > and similarly the OASIS CIQ too.
> >
> > However - you expect that domain analysts can work
> > offline - and then later upload or share their work with
> > both RR servers and co-workers. This is another
> > critical value - that avoids the need for an RR in order
> > to be able to do CCTS. If you use UUID - you are almost
> > mandating that people have to do this with RR - and that
> > clearly breaks the ebXML need for components to be
> > self-sufficient.
> >
> > The good news is - by decoupling away from UUID values
> > and using UID as the reference - any registry can happily
> > import those definitions and assign it own local UUID internal
> > keys to those entries - without breaking anything.
> >
> > Also - by using the UID - you can track groups of related
> > definitions ( query on "USPS0201*" ) and so on - and
> > versions and sub-versions (query on "USPS020120:*") -
> > which is not possible in a UUID based system.
> >
> > Notice also that if you want to use something like UDEF id
> > in addition - or in place of the 6 digit UID - this also
> > works - for the same reasons - the authoritative source
> > creates the UDEF ids and these then become the domain
> > reference system.
> >
> > All these behaviours and more is why the reference system
> > for the semantic building blocks should be using the UID
> > tied to the registry%
>
