Re: 10646, and all that

Dana,
  Perhaps without intending to, I think you have summarized the
arguments against the "Content-type: text/plain; charset=ISO646-
[Han_Language] - [Devanagari_Language]" approach:

(1) "Extension to more languages opens us up to a pandoras box."

(2) "Obviously this would best be used for a message which excluded any
other Han and any other Devanagari content, and it implies that both
would be present, else neither specification would be necessary."

(3) Conversely, like any other per-body-part langauge tagging, it is of
no use for a body part that contains, e.g., more than one Han content.

(4) "Content-type: text/plain; charset=ISO646-Bengali-Japanese
    "Content-type: text/plain; charset=ISO646-Japanese-Bengali"
  I presume that you mean "10646", since, as far as I know, there are
ISO 646 national variations for neither Bengali nor Japanese Kanji.
  I don't know about others, but I have pretty severe problems with
moving from "charset parameter is an unformatted string representing a
registration" to "charset parameters starting in ISO-10646- must be
further parsed and interpreted, starting with the second hyphen, to do
language differentiation".  It seems to me that the latter is much
better handled by either additional parameters or additional header
fields.

    john