Subject: Re: RIM and RS Issue: Identifier Usage (UDDI and URN)
Len Gallagher wrote: > > Michael, > > Thanks for your comments and the time you took to compose them. > > I don't really dispute anything you say -- but I still want the freedom as > a submittor to give URN identifiers to the objects I register. And it's so > cheap to give me that freedom - just a single new attribute in > RegistryEntry and in Organization, the enforcement of an alternate key on > that attribute for RegistryEntry and Organization instances, and the > maintenance of a private, three-column index to map easily back and forth > among UUID's and URN's so you can give me the performance I want no matter > which identifier I choose to use. > > Oh -- I might dispute one thing. I agree that the first <arbitrary > namespace> in a URN is currently meaningless, but that's pretty easy to > rectify. I'd expect "ebxml" and "oasis" and a few other names to be > immediately registered, thereby giving those groups the authority to > further structure their namespaces internally. In fact, we could begin just > by turning internet domain namespaces around for each company, e.g. > > urn:com:xmlglobal > > urn:gov:nist Yes, but people will still invent their own semantics for identifying content if you give them the means. I figured that was what ExternalIdentifiers were for. While we're on the topic of identifiers, can anyone provide a distinction between "name" and "id" in a registry object? Is one composed of the other? Please give examples. I understand that id is implied because the server generates it if the client does not provide it. Does the server also generate a name? Can a name be null in the underlying information model? -- // Michael Joya // XML Global Research and Development // 1818 Cornwall Ave. Suite 9 // Vancouver, Canada // 604-717-1100x230
Powered by
eList eXpress LLC