Re: [URN] draft-ietf-urn-nid-req-01.txt

Dan Connolly (connolly@w3.org)
Fri, 28 Mar 1997 15:00:42 -0600


Message-Id: <333C317A.1BD455E5@w3.org>
Date: Fri, 28 Mar 1997 15:00:42 -0600
From: Dan Connolly <connolly@w3.org>
To: "Karen R. Sollins" <sollins@lcs.mit.edu>
Cc: uri@bunyip.com, ietf-url@imc.org, urn-ietf@bunyip.com
Subject: Re: [URN] draft-ietf-urn-nid-req-01.txt

Karen R. Sollins wrote:
>  Hence there is a certain burden of
> proof on the proposer of an identification (URN) namespace that does
> not hold true for a location (URL) namespace.

Hmm... OK, if there are some process rules that we believe
will result in more persistent, reliable names, then it's
worth writing them down.

I think it's a contractual issue, and not worth specifying in
a technical sense. But that's just my intuition, and in the
case of process, the relevant question is whether the participants
of the IETF are willing to standardize some contractual
properites. It seems that you/they/we are, so away we go...

But it seems perfectly feasible, to me, for the connection
between a URI and this URN process contract to be
outside the identifier itself. That is, rather than

	URN:ISBN:4534-345-345

which syntactically binds the name to the URN process contract,
it seems sufficient to write

	ISBN:23423-234-234

and have the ISBN scheme specification say "this scheme
conforms to the URN process contract."

If not, please somebody explain why not, and write it
up in a spec, and let's review it.

-- 
Dan Connolly, W3C Architecture Domain Lead
<connolly@w3.org> +1 512 310-2971
http://www.w3.org/People/Connolly/
PGP:EDF8 A8E4 F3BB 0F3C FD1B 7BE0 716C FF21