Re: [idn] Proposal for creation of new gTLD for IDN

[Paul Hoffman / IMC <phoffman@imc.org>]

At 10:02 AM +0800 7/7/00, James Seng wrote:
>Back on the basic of Jang's proposal. One of the merits Jang bring to
>the table is that his encoding format are more compact. The reason it is
>able to be smaller because it makes certain presumation on the domain
>names, namely, you dont mix different language in a single label.

Do you mean "language" or "script" here? Many languages are 
represented in multiple scripts. Japanese, for example, is 
represented in three scripts: katakana, hiragana, and kanji (Han). I 
think it would be unfair to the Japanese (and to others who are in 
the same position) to say that they could only use a single script in 
the entire domain name, much less in a single name part. Many 
Japanese company names use a combination of katakana/hiragana plus 
kanji. Similarly, many company names in other languages use ASCII 
digits in their names.

>  It
>furthers compact it by using a base36 rather than the commonly used
>base32.

A huge 10% increase in name length...

>(CIDNUC/RACE compression also make the same presumation, compression
>occurs only when the labels are from the same script 256 blocks so I
>guess you can say CIDNUC also make this presumation indirectly).

Hrmph. It has always made this presumption quite directly and explicitly.

>Yes. One of the problem is that you will not be able to mix Chinese,
>Japanese, Korea, French, German etc in a single label or domain. On the
>other hand, for sanity check, do you *really* want to have such a domain
>name?

Absolutely yes. As a trivial example, www.<chinese characters>.com. 
Many people want to continue to use "www" in order to be consistent 
with the rest of the world. Under the proposal, that name would be 
prohibited and each script would have to come up with its own, new 
equivalent for these types of host names.

A more serious example is an American company with a Chinese 
subsidiary that wants to have a host name with that subsidiary's name 
in it. It is only polite to do so with the best representation of the 
subsidiary's name. Under the proposal, the Americans would be forced 
to continue to be rude and not have the subsidiary's actual name in 
the host name, but instead use a pinyin or Wade Giles transliteration.

As you can tell, I don't believe laying these heavy restrictions for 
a 10% length improvement makes sense. If I am misunderstanding the 
proposal, and it has other merits, it would be good to hear them.

--Paul Hoffman, Director
--Internet Mail Consortium