[DNSOP] Re: [Ext] Re: [EXTERNAL] Re: Call for Adoption: draft-davies-internal-tld

[David Conrad <[e-mail]>]

Joe,

On May 5, 2025, at 10:56 AM, Joe Abley <[e-mail]> wrote:
> On 5 May 2025, at 16:58, Petr Špaček <[e-mail]> wrote:
> 
>> I hear all of you, but do not comprehend.
>> 
>> Could you explain what are the differences between
>> - internal.
>> - home.arpa.
>> - 10.in-addr.arpa.
>> 
>> Please?
> 
> 1. Two of those domain names definitively exist in the public namespace. One does not.
> 2. Two of those domain names exist in a part of the namespace over which the IETF has control and authority; one does not.
> 3. Two of those domain names are implicated in IETF specifications; one is not.

Out of curiosity, do you think https://www.iana.org/assignments/service-names-port-numbers/service-names-port-numbers.xhtml should be limited only to protocols the IETF has "control and authority” and are “implicated in IETF specifications”?

> Another question, what is the difference between:

I’ll play!

> - internal.
> - corp.
> - oweifeowihfoihewfwe.
> - zz.
> 
> Possible answers:
> 1. Three of of those domain names will never exist in the public namespace by virtue of existing policy outside the IETF; one might.

Never is a very long time. Given ICANN’s recent action, it seems likely .internal will never exist in the public DNS.  However, some folks have argued that per RFC 1591, since ISO-3166/MA is seen as authoritative for what will become 2-letter TLDs, it is inappropriate to make assumptions about any 2-letter code.  Also, I’m unaware of any explicit statement that .corp will never be delegated, so this answer appears to be wrong.

> 2. Only two of them are apparently interesting enough to write an internet-draft about.

I can recall drafts about .internal, .corp, and .zz (well, ok, about the ISO-3166/MA reserved 2-letter codes), so this answer appears wrong.

> 3. Nothing, in the sense that none of them require any special handling by DNS servers or clients.

Since this thread has been discussing whether or not there needs to be special handling of .internal (e.g., DNSSEC goop), this answer appears to be wrong.

> 4. Nothing, in the sense that all of them require special handling by DNS servers or clients.

Without more info (e.g., ICANN declares "oweifeowihfoihewfwe“ an alias for ".internal”), I don’t see how “oweifeowihfoihewfwe” would require special handling, so this answer appears wrong.

> 5. Nothing; if the IETF has something to say about .internal, it should also have something to say about all the others.

I don’t see why the IETF should have anything to say about “oweifeowihfoihewfwe”, so this answer appears to be wrong.

My proposed answer:

Other than not being delegated in the public DNS, they are all different:

- One has been proposed for internal DNS scenarios by the party that is generally agreed to be the administrator of the top-level domain name namespace.
- One was used long ago in documentation for internal DNS scenarios by a large software company and is now argued to be too tainted for public use.
- One isn’t particularly special, other than maybe being a bit long.
- One is deemed to be under the authority of ISO-3166/MA due to RFC 1591, so no assumptions can be made about it.

In some cases, resolver operators, out of the goodness of their hearts, may have configured their resolvers to respond locally to queries including one or more of those TLDs because they heard rumors that they may be used for internal use, but they may not be sure where they heard that...

What do I win?

> I can no longer tell whether this is a serious reply or not.

I’m going to go with “not” here.

Regards,
-drc