Re: [Inip-discuss] Domain Names

[Edward Lewis <[e-mail]>]

I'm not sure if that answers the question, but the pointers are helpful.  It
might seem odd that I posted to the list without knowing of this material,
but I didn't get on the list in a conventional way.

In the description on the program page, I see a reference to RFC 6055 "IAB
Thoughts on Encodings for Internationalized Domain Names" (I encourage the
use of document titles and not numbers if you want clarity!).  I don't think
that that document is the best starting point for the discussion.  It's
useful, but not the start.  For one, it's related to one use case - names
rendered in ways not compatible with the DNS (aka IDNs).  IMHO, any thinking
on this topic that starts with the DNS and pushes outward is like the
mindset that Copernicus fought - that is, people believed the Earth was the
center of the universe and tried to explain planetary orbital dynamics as
relying on the Earth as center of the frame of reference.  If we do that
we'll have really ugly formulas to predict where the planets will be seen in
the night sky.

The DNS is an outlying planet in this game.  The DNS supports a subset of
what could be called Domain Names, setting aside that there's never been a
formal definition of Domain Names.  Gathering up the uses of Domain Names in
various applications and protocols that have been reviewed in the IETF,
Domain Names aren't even the only way to create a naming system.  (Prior to
the maturation of the DNS there were many commercial attempts.)

The DNS by itself isn't a naming system.  As a protocol, it's a distributed
database that scales really well and manages to deliver answers quickly.
It's not so hot when it comes to it's data model and other features.  (In
the office yesterday, I used the analogy that one could transport a couch in
a two-seat convertible with the top-down, but there are better tools for the
job.  The DNS can do a lot, but it is strongest in just two areas.)

(It's always bothered me - a DNS name is a name, an address, and a route.
It's a name in the sense that in isolation, it denotes a unique thing.  It's
an address inside the DNS data model tree.  It's a route telling one how to
find the data in the servers, because you always start at the root.  Now in
naming system 'theory', the three are supposed to be different - names are
absolute, addresses are relative to the object's location, and routes are
relative to both the object and the seeker's position.  Somehow in DNS that
all collapsed, maybe now I am beginning to understand why.)

When I began working on the draft called Domain Names, I had to divorce
myself from believing the DNS was the origin of all of this.  After the
early research I realized that the DNS was more like, well maybe, Jupiter
than the Sun in the conversation to be had.

RFC 6055 is an important document in a survey of how identifiers using
non-ASCII characters are rendered into protocols that assume ASCII.  From an
architectural layering perspective, how a name is converted into the binary
transmissions is way down in the stack.  (When I wrote my draft and early
review convinced me to not go down that far in such a stack, which is why
this sticks out to me.)  I don't have a well cooked layer of stacks in mind
but I would say that figuring out how to convert the identifier from a name
to something useful (like a network address), that is, how to resolve the
name, is more significant.

I mentioned before the draft Domain Names.
https://tools.ietf.org/html/draft-lewis-domain-names-01  The draft is
incomplete because, "there's a conversation to be had."

On 12/18/15, 14:22, "Inip-discuss on behalf of Warren Kumari"
<[e-mail] on behalf of [e-mail]> wrote:

> You mean this?: 
> "NIP-discuss is the public discussion list for Internet Names and Identifiers
> Program. See the full description at
> https://www.iab.org/activities/programs/names-and-identifiers-program/
> 
> The inip-discuss list is an open-subscription list, sponsored by the IAB,
> intended to be an open venue for discussion of the topics covered in the scope
> description for the IAB Internet Names and Identifiers Program
> (https://www.iab.org/activities/programs/names-and-identifiers-program/).
> These topics broadly include the use and evolution of naming systems for the
> Internet, including but not limited to DNS and other protocols that use domain
> names. Contributions will be covered by the IETF Note Well
> (https://www.ietf.org/about/note-well.html).
> 
> The list is intended to provide input and perspective, at the architectural
> level, to discussions in current or potential IETF, IAB, or IRTF activities on
> the topics the Program was formed to address. It is not intended to replace or
> supersede discussions in the IETF on specific protocol matters, such as those
> addressed in the charter for the DNSOP WG. It is also not intended to replace
> or supersede having a formal membership and deliverables for the IAB program,
> which is managed at the discretion of the IAB."
> https://www.iab.org/mailman/listinfo/inip-discuss
> 
> 
> The IAB page says (amongst other things):
> "The program seeks to provide a useful framework for thinking about naming and
> resolution issues for the internet in general, and to deliver recommendations
> for future naming or resolution systems.
> 
> The scope of initial investigations is deliberately somewhat open, but could
> include:
> 
> some basic terminology: what do we mean by "names," "identifiers," and "name
> resolution" in the internet? What attributes of naming systems and identifiers
> are important with regards to comparison, search, human accessibility,
> overview: where are naming protocols and infrastructure important to the work
> of the IETF (and perhaps elsewhere)? Where is the DNS being used (and perhaps
> stretched too far)? What other identifier systems are we coming up with, and
> how well are those efforts working? This area will include examination of some
> of the naming systems under development or in use elsewhere, such as NDN, as a
> way of informing our thinking.
> For protocols (inside the IETF or outside), what should protocol designers
> know about re-using existing naming systems or inventing their own? Are there
> guidelines we can usefully provide?"
> 
> Which leads to some of my confusions about how this related to the design team
> vs discussions in dnsop, what the status of all this is, etc.
>  
> W
> 
> 
> 
> 
> 
> On Fri, Dec 18, 2015 at 10:45 AM Gihan Dias <[e-mail]> wrote:
>> On 2015-12-18 පෙ.ව. 2:34, Warren Kumari wrote:
>>> >
>>> > So far yours is the only message on the list (other than the
>>> > "welcome") - it's still fairly unclear to me what the relationship is
>>> > between this list and the IAB program (is this a write only list? What
>>> > is it supposed to be commenting on / what output is there from the
>>> > inip program (if any), what is the relationship between that and the
>>> > design team? Etc)
>> I too was waiting for the list to get populated and then active.
>> 
>> Could someone post the list ToR?
>> 
>> Gihan
>> 
>> _______________________________________________
>> Inip-discuss mailing list
>> [e-mail]
>> https://www.iab.org/mailman/listinfo/inip-discuss