Re: [idn] An ignorant question about TC<-> SC

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

Deng:

Thank you for the two analyses. This is a good basis for discussion 
for the users of many languages, not just Chinese.

At 10:37 AM +0800 10/30/01, xiang deng wrote:
>1.Adopting multiple records solution - analysis
>If a multiple records solution are adopted, and all possible
>TC/SC forms of a single Chinese domain name are added into a
>zone file as the different forms of one identical domain name,
>then this approach has the following weakness:

Why did you assume that all forms must be added? I thought it was 
generally agreed that a name that consisted of, say, four traditional 
characters would probably only have two forms (all-traditional, and 
all-simplified). A small number of names would have one or two more 
(maybe TTTS or STSS), but that the number would almost always be very 
much less that 2^n.

>(1)2^n problem. To ensure a Chinese domain name both TC/SC to
>be correctly resolved, 2^n records must be registered and all
>the records point to one IP address (Note: n indicates the total
>number of the TC/SC variants of the domain name). Chinese has
>over 4,000 variants, and most of them are in common use, a single
>ordinary domain name may have dozens of, hundreds of even thousands
>of TC/SC records, if all of them are to be registered, then total
>number of registrations will explode. In order to protect their
>domain name, users must register 2^n domain names, registration
>cost will increase rapidly, and this is obviously not practical.

Why is there any cost at all? If there is general agreement that all 
sensible spellings should be registered at once, why should a 
registry be able to charge by the record instead of by the name?

>(2)The lower level delegation domain name servers may adopt a
>different domain administrative policy or they simply don't support
>multiple records registration, therefore the consistency of TC/SC
>domain names cannot be ensured.

Exactly right. This is precisely the same problem faced in every 
language that has spelling variants or meaning variants. In fact, it 
also appears when company A buys company B and they try to run their 
domains in parallel.

>(3)How to ensure that the 2^n records are to be owned by a single
>registered user? This is also an inevitable question, if this cannot
>be ensured, there are will be a large number of domain name
>disputationes. Different registration authorities has different
>registration policies, even if the registration policies can prevent
>gTLD and ccTLD practice; they don't have control over secondary or
>tertiary domain.

Indeed. A single registry can ensure that all spellings of a name are 
held together in the one zone, but they cannot assure anything in a 
lower zone, and cannot assure anything in a different registry. 
Again, these problems are the same for almost every language.

>2.Adopting TC only records and SC only records solution - analysis
>TC only records and SC only records solution has the following weakness:
>(1)Not applicable to Hong Kong and Macao where both TC and SC are used,
>simplified Chinese is not used at all in Taiwan, TC only or SC only
>records cannot meet the requirements of the users.
>
>(2)This approach cannot solve technically delegation domain problems either.
>
>(3) How to ensure that the two kinds of records are to be owned by a
>single registered user? This is also an inevitable question, if this
>cannot be ensured, there are will be a large number of domain name
>disputationes. Different registration authorities has different
>registration policies, even if the registration policies can prevent
>gTLD and ccTLD practice; they don't have control over secondary or
>tertiary domain.

This sounds exactly right, and problems 2 and 3 match those above.

--Paul Hoffman, Director
--Internet Mail Consortium