Re: [Simple] Address and person URIs

[Henning Schulzrinne <hgs at cs.columbia.edu>]

> To be clear, is your intent here that urn:service would be defined as 
> allowing registered values through an IANA registry or something? i.e., 
> if I come up with a new service, foo, I could register it and thus 
> urn:service:foo would be defined?

Yes.

> I'm not sure how one would properly define the scope of "service" in 
> such a context...

Depends on what you mean by "scope". If you mean "what does the foo 
service really mean, i.e., what do I get", I'd punt: the registrant 
defines the meaning of the service.

I had one simple model in mind: Define by resolution. Each service is 
associated with a DNS resolving hierarchy, e.g.,

urn:service:sos.arpa

is the service defined by the DNS resolution mechanism sos.arpa. The 
assumption is that any proxy that resolves this domain has to know what 
request information is necessary to do the resolution, e.g., by location 
or other request property.

This doesn't necessarily fit the 'real human being' one all that well, 
but some other services, including

urn:service:pizzahut
urn:service:directory-service

work reasonably well, as long as each service can agree on a resolution 
mechanism. This is somewhat akin to an IP-level anycast service: "I want 
any suitable instance of this service and you figure out what that might 
be".

This is obviously very rough thinking at this stage. As has been pointed 
out in private email, it is not clear that the URN definition matches 
this particularly well. However, one could argue that something like an 
ISBN URN has somewhat similar properties, albeit with a much more 
precise contract as to what one might get back.

Henning

_______________________________________________
Simple mailing list
Simple at ietf.org
https://www1.ietf.org/mailman/listinfo/simple