Re: [urn] Registration requirement for separation of namespaces

I'll say this as a participant only, but informed by my experience
over the last four years as the AD responsible for handling URN NID
requests:

Essentially all (perhaps exactly all, though I can't remember
definitively) NID requests that I've handled over the last four years
are from organizations who want namespaces for use by their
organizations or industries.  It's quite clear that there is little to
no value in asking them to document why they need a new namespace:
it's because they want one associated with their organization or
industry, and that they control.

There could be situations where it would be useful to ask the
question: if there's already a namespace for, say, naming movies (call
it "movie:"), and a film-industry organization asked for one called
"film:".  We'd definitely want to know why "movie:" doesn't suffice.
The problem is that in order for us to ask the question, *we* would
have to know that "movie:" exists and seems related, and all the
requestor would have to do is to put a boilerplate "we need our own
industry namespace that we control" statement in.  Unless we notice
that they've missed (or intentionally ignored) the related namespace,
nothing will be accomplished.

I'd support something that said that if there are related namespaces,
the request should note them and explain why they are not adequate and
this one is different.  But I'd not support its being a requirement,
because in most cases it's unnecessary and in the others its a
requirement that's unlikely to be enforceable.

Barry, as participant

On Fri, Mar 18, 2016 at 7:03 PM, John C Klensin <john-ietf@jck.com> wrote:
> Hi.
>
> This note is prompted by a recent Last Call discussion on the
> IETF List about a new namespace proposal,
> draft-martin-urn-globus.   Those who are interested in details
> should have a look at that draft and archives of the IETF
> discussion list for messages with subject lines containing that
> string and posted within the last week or so.  I don't believe
> those details are needed to understand what follows.
>
> Section 4.3 of RFC 3406 requires that a registration request for
> a new namespace have a "Namespace Considerations" section that
>
>         "outlines the perceived need for a new namespace (i.e.,
>         where existing namespaces fall short of the proposer's
>         requirements)."
>
> That requirement was dropped when we folded the registration
> procedures and requirements into 2141bis and, indeed, was
> dropped before the posting of the now-obsolete
> draft-ietf-urnbis-rfc3406bis-urn-ns-reg-09 in February 2014 (and
> probably earlier).
>
> That was some time ago.  Just to prevent any last-minute
> surprises, does anyone see a need for a requirement that
> proposals for new namespaces include consideration of existing
> ones and why they are not applicable?  I note that such a
> requirement may require significant research, especially as more
> NIDs are registered, and may not actually be helpful.  I assume
> that is the reason why the requirement was dropped.   If anyone
> does feel that the requirement (or something like it) should be
> reinstated, I'd appreciate it if they would speak up soon... and
> that their comment justify the additional costs and explain how
> whatever claims are made will be evaluated (or not).  Sending
> proposed text would be ideal, but the most important thing, IMO,
> is to know whether that is an issue that anyone plans to bring
> up during WG or IETF Last Call.
>
> thanks,
>    john
>
> _______________________________________________
> urn mailing list
> urn@ietf.org
> https://www.ietf.org/mailman/listinfo/urn