Re: [DNSOP] Interim DNSOP WG meeting on Special Use Names: some reading material

[Andrew Sullivan <[e-mail]>]

On Fri, May 08, 2015 at 08:10:41PM -0700, Paul Hoffman wrote:
> 
> - Will the IETF require some specific metrics for RFC 6761 reservations?
> 
>   - If yes, what are those metrics?
> 
>   - If no, who makes the non-specific decision?

Increasingly it strikes me that RFC 6761 is trying to do too much.

It was, as we all know, created as a _post hoc_ rule to permit the
creation of local, on which Apple squatted to create Bonjour.  That
was one particular use of a namespace: the namespace amounts to a
protocol-shifting mechanism.  We can expect to see this again (and
indeed, several of the names that are in the p2pnames draft are
effectively in this class).

In order to create that set of rules, however, the special use names
registry came about.  That registry includes all sorts of other
special-use names that are in fact reserved for _policy_ reasons.  For
instance, example, test, and invalid are there because of the policy
of needing certain names for testing, documentation, and so on.
Similar arguments can be made about the RFC1918 addresses in arpa:
these are special only because of a different policy decision to
reserve chunks of v4 space for local use.

The interesting case is localhost, which I think might have elements
of each of these properties, though because of the binding to the
127.0.0 network you might be able to argue that the policy is actually
elsewhere.

It seems to me that making new reservations solely on _policy_ grounds
is overstepping our role, because we actually gave that management
function away to someone else many years ago.  But if there are
additional protocol-shift registrations, it would be appropriate to do
that.

This makes me think that what we ought to offer ICANN is a mechanism
to make insertions into the special-names registry by different
criteria than the protocol-shift cases.  The latter all fit neatly
into 6761's "7 questions", but policy-based ones sort of don't.

Anyway, that's a suggestion.

A

-- 
Andrew Sullivan
[e-mail]