[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [idn] call for comments for REORDERING




----- Original Message ----- 
From: "Martin Duerst" <duerst@w3.org>
To: "Soobok Lee" <lsb@postel.co.kr>; <DougEwell2@cs.com>; <idn@ops.ietf.org>
Cc: <jseng@pobox.org.sg>
Sent: Friday, October 19, 2001 5:12 PM
Subject: Re: [idn] call for comments for REORDERING


> At 15:31 01/10/19 +0900, Soobok Lee wrote:
> 
> >Second) If we take into consideration the additional length of subdomains of a
> >IDN, the total length of  <ML-subdomain>.<ML>.com  often  exceeds
> >14. ANd if we think about mailbox name in ML@Sub-ML.ML.com, it will exceed 20.
> >IN that case, the Sum of the lengths of  saved characters in each 
> >substring in ML@SUB-ML.ML.com will be great, around 15~20, i guess.
> 
> It is a very long standing lesson when working with computers
> that in case there is a tradeoff between a more stable and
> a more compact solution, it's in most cases, and definitely
> in the long term, much better to take the more stable and
> simpler one. The Y2K problem is the best example, but there
> are many other ones.
> 

Your menttion of Stability are presume the variations of frequency
distribtion of characters in a script block. But, as i stated before,
the influence is marginal and REORDERING is not optiomal solution,
but sub-optimal solution that have clear benefits for a long time.

Please read  2500-years-old chinese classic books, then you can 
find there has been no great change in frequent han characters.
Really!!!

You should suggest the proof that the frequency
change will be great in the future and hurt the statbility of
REOERDERING tables.


> 
> >We should not attempt to draw an upper limit on future  applications of
> >han/hangeul IDN domains by imposing unfair disadvantage on max length
> >of Han/hangeul labels .
> 
> I don't see any unfair disadvantage. Both in Latin and in Han/Hangul,
> it's already possible to make names longer than really useful.

You missed what the above examples really shows.
Please think again the sum of length of saved characters in
ML@SUB-ML.ML.com.


> To say that overly long names have been cut off at a different
> point for Latin and for Han or Hangul may be true, but it is
> rather irrelevant. And even with the most advanced compression,
> we will never be able to make sure that we are completely fair
> to everybody.
> 
> 
> Regards,   Martin.
>