ietf-822
[Top] [All Lists]

Re: FYI: BOF on Internationalized Email Addresses (IEA)

2003-10-30 22:42:22

At 23:24 30/10/03, Markus Stumpf wrote:
A big fair in Munich had tested kinda electronic cards lately. As you
buy your ticket you type in your contact data and it is printed on the
card as a barcode. Exhibitors had barcode readers and special
software, so if you want to make a contact you hand over your ticket
and the infos flow ...

Markus,
We are here in a typical layer violation we do not see. We think that passing information from layer 7 to a user is a simple task because ascii simplifies it to ascii readers. ENS Model puts 2 layers for that.

Your example is a very good common example of this:
- the application layer (7) has an infromation: the visitor carries and delivers his name to the registration system - the interapplication layer (8) corresponds to the trade show system which organizes and manages the bar codes or other systems (magnetic, noise printing, laser, etc.) and the event and the lanes for him to stop at your boot using his feet and not bytes on a wire. - the assistant layer (9) which is going to read the information for you and/or for your computer. - until eventually you the user (layer 10) may access to the infromation and possibly disseminate to other layer 10 colleagues or colleges, or to escalate it to other communicating structures (11). For example the visitor may be a common client for your trade pavillon and benefit from a discount. And then you may want to verify his legal ID and hit the societal system layer (12).

All these layers car be executed by people and by machines. The simplest protocol is yes/no. It then increases in complexity through numbers, then langages and characters. Basic English and ASCII are one of the most interesting level for naked brainq. But what about computer (software) or community assisted (brainware) brains: whre is/are the most interesting levels of equilibrium?

This is why I say the target is not to develop internationalized, nor multilingual names. To please Hindi vs American (objections stand). But a truly fully vernacular support for users and applications. This means even more than transparent binary, because binary is only to be one occurence in it.

We have a default : ascii. We have a first level of complexity : IDNA. This must continue. This is why I objected to the (xn--) structure. It is to be necessarily the begining of an unlimited vernacularisation. Having entered the xn-- in the left of the name label introduced a cross hierarchical structure in the DNS (levels from right to left and layers within the labels from left to right). With a complex impact on the DNS administration and evolution - as we start seeing it now. Another exemple is IPv6. I doubt we can continue writing manually db.flles addresses.

Now, several question the need of a change. IMHO they are right and wrong as we do not consider the layers. At the user's layer we must have everything in the language and culture of the user. We are not going to say that cultures are bugs to correct to fit IDNA! But at the old SMTP level or DNS do we need that complexity? I think not. Even if we were building a brand nwe DNS.

What Vint suggests is to use ASCII as a 35 or 37 highly readable numbering pad (2 to 9 A to Z and "-", or 0 to 9) to give more capacities to support vernacularization on the layer above. Otherwise everything is going to collapse into a too complex system. For example where are you going to introduce the OPES which will read the written chinese in the proper local way? This is the same problem as for the multilingual TLDs. In a flat model we saw it was not possible. I do not see a real problem in a properly set multi-layer model. One layer can go one way and the next layer the other way. Then you know what you wand to do where.

jfc












There WILL be a lot of problems with internationalized email addresses
but IMHO a lot of them will be solvable by sophisticated software.
Some of them won't as in print media (and also tv. radio, ...) a domain
    µ.de        (in 8859-1)
and
    µ.de        (in 8859-7)
looks identical, but isn't as the µ sit on different codes. So the
commercials will be fun:
    "visit us in the Internet on www.µ.de in good old 8859-1 coding."

And despite all these I think it will be a long time until addresses like
    jürgen(_dot_)größl(_at_)hüberbräu(_dot_)de
will look familiar (even if this is a (for me) local coding).

        \Maex

--
SpaceNet AG            | Joseph-Dollinger-Bogen 14 | Fon: +49 (89) 32356-0
Research & Development |       D-80807 Muenchen    | Fax: +49 (89) 32356-299
"The security, stability and reliability of a computer system is reciprocally
 proportional to the amount of vacuity between the ears of the admin"


<Prev in Thread] Current Thread [Next in Thread>