Re: [idn] opting out of SC/TC equivalence

[=?utf-8?B?dHNlbmdsbUDoqIjntrLkuK3lv4Mu5Lit5aSnLnR3?= <tsenglm@cc.ncu.edu.tw>]

Hi !  Liana:
        I have the same feeling . We try  to solve these SC/TC problem step
by step and  to reduce the complexity.  As John had pointed-out in this
listing , sometimes many scripts-only term  will be expand to lanaguage-term
and let some one to neglect the key-point.  1-1 mapping will reduce the
complexity like mixing case of alphabet that will reduce the number of
non-necessary registration and confusing of using.
        No one try to translate TC/SC in DNS. Even an expert of chinese
language also can not resolve  the different local naming  and habit of
personal name in different Province of China.  The PRC-SC is the frequently
used characters in chinese language, some of them can be differentiated by
different shape, but a little of them is easy-to-confuse ,  such as TC(頻)
SC(频)  SC (统)  TC (統). James seems avoid to answer what is the first step
to solution.
        If the easy-to-confuse characters can be reduced . Even a little ,
that is better than nothing to do.  Reducing the number of easy-to-confuse
characters in nameprep is a better help to CJK users.
        Always to point out the complexity of chinese language and never to
solve some problem step by step is not a coorect approach in engineering.

L.M.Tseng
----- Original Message -----
From: <liana.ydisg@juno.com>
To: <James@Seng.cc>
Cc: <idn@ops.ietf.org>
Sent: Friday, August 31, 2001 4:02 PM
Subject: Re: [idn] opting out of SC/TC equivalence


> Hi, James,
>
>   That was a miscommunication on my part, that I did not
> amend, since I did not see anyone care about except
> Martin, which I have answered indirectly.  This is
> realy two issues. One is one-to-one TC/SC mapping
> in [nameprep]. With this part, we shall include GB, Big5,
> and a few other standards representing local access
> directly to [nameprep] and speeding up hostname
> translation.  This is my 20% effort for 80% TC/SC cases
> arguing for. [nameprep] shall only do one-to-one
> mapping, no guessing or second try shall be there at
> all.
>
> Another issue is direct mapping to StepCode part of
> the proposal which is mixed up in that short
> communication.  StepCode has Pinyin for each radical
> already for SC character set.   More work is needed
> to check against with larger character set such as Big5
> and Unicode.   I thought if some published listing may
> help out for Kanji and Hanja to encode,  just an idea to
> throw out to be criticized, it does not mean to play with
> TC/SC AI treatment in [nameprep].   After that, I was a
> little regret, since  if the current architecture of [nameprep]
> does not change,  I am putting up more work for nothing.
>
> Liana
>
> On Fri, 31 Aug 2001 12:50:45 +0800 "James Seng/Personal" <James@Seng.cc>
> writes:
> > Iiana,
> >
> > I am not comment on other parts but your presumation that TC/SC can
> > be
> > done by radical equivalence simplification is unfortunately
> > misplaced.
> > No doubts there are *some* cases whereby TC->SC can be done by
> > changing
> > radicals, it is not rule of the thumb.
> >
> > Once again, you may want to read the problems with TC-SC paper by
> > Jack
> > Halpern at http://www.basistech.com/articles/C2C.html first on basic
> > difficulties of TC/SC in general.
> > http://playground.i-dns.net/one/onec_sum.htm (also by Jack)
> > describes
> > the more specific difficulties of doing TC-SC in Domain Names.
> >
> > -James Seng
> >
> > > As to TC/SC, I think the problem can be divided into two
> > > levels. One level is the mechanical, one-to-one.  They are
> > > the majority and bothering  the readers.  The minority as many
> > > have said those n-to-1problem, normally one of them is
> > > the major one, and the other are minor ones.  There are
> > > always exceptions!!!  So to benefit the majority, the way to deal
> > > with it is to come up with an arbitary one-to-one, and let Chinese
> > > to fight which one is the major one!   The same principle is
> > > applied to decomposition of a Hanja and Kanji.  When there
> > > is a base, the rest will follow , the application will have
> > > something to conform to.
> >
> >
>