Re: [idn] Interesting links

[hoho <hoho@iis.sinica.edu.tw>]

Sorry, Doug,

When you mentioned the two numbers 4 and 6, I think you are talking about the
average number of characters and average number of characters with variants per
label. However,  a domain name is composed of several labels. Let's consider your

example. Then, a domain name would contain more than 10 characters each with at
least two different codepoints in Unicode as long as it contains 3 labels. You
see,
this is exactly the case in which we have more than 1,024 equivalent Chinese
domain
names.

-- Janming


DougEwell2@cs.com ¼g¤J¡G

> In a message dated 2002-02-12 9:50:11 Pacific Standard Time,
> hoho@iis.sinica.edu.tw writes:
>
> > Dear Dough,
>
> Hmm, I guess it really is difficult to type names correctly.  :-)
>
> > As several members of the list tried to explain to the group, it is
> > very difficult to ask a user to enter a Chinese domain names exactly
> > in a Unicode input environment. This is because the complexity of
> > getting the exact domain name is exponential in the length of a
> > domain name. In other words, if there are 10 characters with variants
> > in a domain name, then there are at least 1,024 different Unicode
> > strings corresponding to the same name. It is frustrated for users to
> > deal with such a high complexity.
>
> Continuing to talk about 1,024 variants is unrealistic and is causing you to
> lose credibility, because
>
> (1) the vast majority of Chinese domain names (in all surveys and in existing
> testbeds) are, and will be, much shorter than 10 Han characters, and
>
> (2) a significant percentage of Han characters do not exist in TC/SC pairs.
>
> If you consider a CDN that is, much more realistically, 6 characters in
> length, and 4 of those 6 characters can be expressed as either TC or SC, then
> you have 2^4 = 16 possible variants.  Now, it may well be that registering 16
> names is an unreasonable burden.  I am not saying that there is no problem at
> all.  But it does no good to exaggerate the problem by claiming that it will
> cause THOUSANDS of variants of a typical domain name.
>
> -Doug Ewell
>  Fullerton, California
>  (address will soon change to dewell at adelphia dot net)