Re: [urn] call for comments: an alternative 2141bis document

[Keith Moore <[e-mail]>]

On 02/19/2013 11:13 PM, Peter Saint-Andre wrote:
>
> Agreed.
>
> And to Larry's original point, RFC 2141 says only that URNs) are
> *intended* to serve as persistent, location-independent resource
> identifiers. It doesn't guarantee that those intentions will be
> achieved in reality.
>
> Here's my interpretation (which doesn't necessarily take into account
> the history of discussion on this topic because I wasn't involved back
> then and I lack the full context that Larry, Ted, Leslie, et al. possess):
>
> 1. Persistent doesn't mean permanent (since nothing is permanent), it
> only means long-lived (or longer-lived than a location-dependent
> identifier).
Actually the binding between a URN and the resource, once assigned, was 
indeed intended to be permanent.  That doesn't mean that the resource is 
permanently accessible (or ever accessible).   And it was always 
understood that the resource could change (in the sense of being 
replaced with a later version of the same thing).   But the URN should 
never be reassigned to a resource that is inconsistent with the original 
binding.
> 2. Location-independent means the resource is not tied to a particular
> path at a particular DNS domain name via a particular access method.

It means all of that and more.   For example, it also means that the 
meaning of the URN is the same from every location in the network.
> IMHO, persistence and location-independence are closely intertwined.

Yes, location-independence is a necessary condition for persistence.
> I also think that it's good to have identifiers that are closer to
> being persistent and location-independent (or farther from being
> ephemeral and location-dependent), and I think that URNs as defined in
> the existing specs fit the bill. All we're trying to do here is
> modernize the definitions a bit to bring them in line with RFC 3986,
> RFC 5226, etc.

Of course I realize that no protocol specification can force any 
particular URN to be either persistent or location-independent 
forever.   But that's not a justification to revise the URN 
specifications to dilute the intent of URNs.

Keith