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

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



Hi Patrik,

Unfortunately, I am not as technically knowledgable as you are so I
rather not comment on what you have wrote.  However, what I am certain
is that you have illustrated the fact that
applications/clients/users/servers/etc can be made to take advantage
of this explicit labeling of what "script" an IDN is in and over time
(with people writing appropriate applications) can be developed into a
very powerful and useful system.  (Unlike TLD such as ".gov", ".ca"
which serves next to no purpose from an IDN's perspective.)

Thanks,
Ben


----- Original Message -----
From: "Patrik Fältström" <paf@cisco.com>
To: "ben" <ben@cc-www.com>; <DougEwell2@cs.com>; <idn@ops.ietf.org>
Cc: <duerst@w3.org>; <klensin@jck.com>
Sent: Friday, October 26, 2001 1:32 PM
Subject: Re: [idn] An ignorant question about TC<-> SC


> --On 2001-10-26 12.48 -0400 ben <ben@cc-www.com> wrote:
>
> > Thanks for pointing out the confusion.  Perhaps just the word
"script"
> > is better.  To clarify what I mean, it is to label a Chinese
domain
> > name as <idn>.<traditional> or <idn>.<simplified> (and not
> > <idn>.<chinese> as this serves no purpose) inorder to take out the
> > guess work for the CDN users.
>
> This implies that the DNS server need to know how to do matchings
between
> different charsets (I switch term), for example between BIG5 and
some
> version of GB, or between ISO-8859-1 and ISO-8859-11. Given some
1000
> charsets, the server need to have huge equivalence tables, because
in DNS,
> we only do exact matchings.
>
> Instead of asking the server to do all of this work, one can force
the
> client to translate the local charset to one and only one. The
server then
> only need to know how to match within that charset. And, the client
only
> need to know how to map from the local charset to the unified one.
>
> We then took one more step and asked the clients to also do
normalisation,
> so the servers can do matchings based on direct comparison on the
bits
> which is the representation of the domain name itself.
>
>    paf
>
>
>