Re: [DNSOP] WGLC for draft-ietf-dnsop-alt-tld

[Brian Dickson <[e-mail]>]

On Mon, Apr 3, 2017 at 7:37 PM, George Michaelson <[e-mail]> wrote:

> I think that's a useful mail. So in that sense, I have a question:
> Would you say anything to this, were you in edit mode, on a draft
> going to LC if that draft didn't say it?
>
> If you had a draft requesting a TLD to "exist" in some sense: in or
> not in a registry; passed or not passed into the DNS; delegated or not
> delegated via ICANN; would you reference the IAB document? If not, why
> not?
>

Whether to reference the IAB document?
In DNS - no, since the draft would conflict with the IAB statement. It
would open the ICANN-IETF can of worms, which the IAB document suggests
(directs?) we not do.
Delegated or not, doesn't come into it I don't think.

Not in DNS, but in a registry? Maybe, but only to make some distinction
between what the IAB document says, and some compelling reason.

Honestly, I think the IAB statement should appear pretty much as widely as
possible.
It should be up front on the DNSOP WG page.
If it could be added to "note well", that would be cool, too.
If there was a URL shortener, make the shortened URL one of the IETF
wireless SSIDs! :-)

As far as any TLD in the DNS is concerned -- there are only two basic ways
that can explicitly happen with a signed root - a delegation (signed or
unsigned), or a DNAME.
I don't think there is a third way, except maybe with specific RRTYPE(s) in
the root zone, for which I don't think a defined mechanism exists,
procedure-wise.

Brian



>
> -George
>
> On Tue, Apr 4, 2017 at 11:26 AM, Brian Dickson
> <[e-mail]> wrote:
> > In response to the latest comments by Paul Hoffman and George Michaelson,
> > I'd like to offer my $0.02 on the meaning and purpose of the alt TLD vs
> the
> > IAB statement.
> >
> > My read is (whether or not it is correct) that there are three
> possibilities
> > for a special name.
> >
> > The first is, a special but needs DNS resolution. This is one case the
> IAB
> > says, "register it and put it in DNS under arpa". (I don't think that is
> > controversial at all, and a wise recommendation.)
> >
> > The second is, a Very Special, but does not belong in DNS.  (IAB second
> > option.)
> >
> > The third is, a Not Very Special, and not in DNS. Not registered, FCFS.
> Not
> > covered by the IAB statement by virtue of not being registered, but IMHO
> not
> > conflicting with the IAB statement.
> >
> > Very Special: It gets its entry in the registry in order to establish its
> > uniqueness, but isn't in DNS, so no entry under arpa. This avoids the
> > possibility of multiple mechanisms for interception fighting with each
> > other, since the behavior is (or should be) name-driven. Also wise, and
> also
> > in-scope for the IAB statement.
> >
> > Not Very Special: whoever wants the name, is reasonably sure it won't be
> > exposed outside of a closed environment (e.g. a single application), and
> > doesn't want or need to go through the 6761 process to get the name
> > registered.
> >
> > Not Very Special is basically 6761 without the registry, in a first-come,
> > first-served, no guarantees kind of way.
> >
> > The "onion" thing showed the need for some way of avoiding TLDs, avoiding
> > conflicting names, and avoiding heavy process, IMHO. And I think "alt" is
> > the right answer.
> >
> > Also IMHO, making it "alt.arpa" would be very confusing; I think any time
> > someone sees "arpa" as the TLD, they should believe it exists in the DNS.
> >
> > Having "alt" be the parent name here, and not be in the DNS, keeps things
> > clear even to non-DNS folks.
> >
> > And finally, maybe there is a use case for FCFS local-use names that
> kind-of
> > are in the DNS. If such a need were to arise, then THAT would be
> something
> > where "alt.arpa" would make sense. But given the relative ease in adding
> > things under arpa, I don't see a good reason for creating non-registered
> > FCFS when registered FCFS is available, under arpa.
> >
> > Brian
> >
> > _______________________________________________
> > DNSOP mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/dnsop
> >
>