Re: Proposed change to draft-kindberg-tag-uri: tags and "newsml" URNs

Isn't it OK that the owner of the IP number implicit in 
e7fe1b91-8cd5-0310-98dd-2f12e793c5e8 might not be able to say what that 
particular UUID was used for? That doesn't affect the usefulness of this 
particular ID as a repository identifier.

Or take the Atom blog entry with an id of,1999:blog-3063265 -- is there something wrong because might not have a direct way of knowing exactly what it was 
the id of?

Of course, there is a social protocol at work here (how could one have a 
naming system without one?); e.g. if someone minted tags beginning,2005-03-04: without my permission then there 
could be problems -- just as they shouldn't mint UUIDs from my machine's 
IP number.

If "denotational accountability" was an issue in a particular 
application context then I would expect a protocol to arise for dealing 
with that. Tags don't *have* to be used opaquely.



John Cowan wrote:
> Larry Masinter scripsit:
>>Yes, there are two separate namespaces that both use
>>dates and named "name minters", but urn:newsml actually
>>identify resources, and the nature of the resource identified
>>can be determined by asking the Provider identified with
>>the ProviderId,
> (Identified at the date specified by the DateId, provided that Provider
> is still extant.)
>>while with 'tag', there is no authority to ask, and all of the semantics
>>are inferred from the context of use.
> So if I ask a tag minter at "What did you mean by <,
> 2005:whatever>?", it's perfectly in order for the tag minter to reply
> "I have no clue"?
> This is a social, not a protocol, question.


Tim Kindberg
hewlett-packard laboratories
filton road
stoke gifford
bristol bs34 8qz
voice +44 (0)117 312 9920
fax +44 (0)117 312 8003

Received on Friday, 4 March 2005 07:29:09 UTC