Re: [idn] Requirements I-D

["Eugene M. Kim" <ab@astralblue.com>]

On Mon, 15 May 2000, Paul Hoffman / IMC wrote:

| I would be perfectly happy with the first choice. Does anyone here 
| feel that the eventual IDN might have multiple charsets? If so, 
| speaking up now would be good.

To me it seems most critical to keep the DNS interoperable and
homogeneous as much possible so every machine on the Internet must be
able to address each other using a simple method.  This is particularly
important in the world where the DNS will be a primary means to
consistently identify a host, e.g. in IPv4 DHCP/PPP and the whole IPv6
network.

Thus IMHO we must avoid supporting multiple charsets if it leads to an
ultimate chaos where some implementations can become incompatible to
each other due to unsupported charsets.  And that chaos is all too easy
to happen.

Supporting multiple charsets has an issue whether to designate a single
charset as the `canonical' one, say Unicode.

We would be in an obvious chaos if we chose not to do that.  One facet
of the chaos would be how to define a N*(N-1) mapping rules given N
charsets, including how to represent some nonexistent characters from a
foreign charset.  Otherwise we would easily be fragmenting the DNS
space, which is unacceptible.

And it seems that there is no clean way to encode foreign characters in
general *along with their charsets*.

Even if we did designate a canonical charset (which I firmly believe
will be Unicode) there would be a problem that a round-trip mapping is
still impossible, not because of case folding but simply because in some
charsets there are several characters that has no 1-to-1 equivalent in
Unicode.

(In fact many of these characters are pretty silly to include in a host
name so one might say we should define a subset of a given local charset
where every character has its equivalent in Unicode, but defining such a
subset once and for all is a real PITA and unrealistic.  And there is
obviously no guarantee that no crucial characters in a local charset
will be omitted in future versions of Unicode; AFAIK local authorities
can always add such characters into their charsets if necessary.)

Regards,
Eugene

-- 
Eugene M. Kim <ab@astralblue.com>

"Is your music unpopular?  Make it popular; make music
which people like, or make people who like your music."