Re: UTF8

[nisse%lysator.liu.se@localhost (Niels Möller)]

Joseph Galbraith <galb-list%vandyke.com@localhost> writes:

> I suppose if we really want to solve this problem, we do the following,
> we could introduce a new set of authentication messages where the
> client sends BOTH the UTF-8 and the raw data entered by the user.
> Then the server can pick which to use.  The client knows the users
> LC_TYPE, and can therefore safely translate to UTF-8.

The proposal is mostly equivalent to having the client tell the server
what the client's local character set is, which might help servers
with no clue about the user's preferred character set. Such a message
from the client to the server would mean

  "Hey! I don't think you have any clue what character set is used for
   my account. But I know that I prefer koi8r, so please convert the
   data I provide into koi8r before you try to look me up in your user
   database."

Except that if we also send the octets, no actual conversion is needed
on the server side.

I think Joseph's idea is pretty good. I don't want to add it to the
core drafts, but interested parties could surely write up a separate
draft to experiment with.

It would also be possible to configure a client running on utf8 to
work with an "agnostic" server with koi8r (say) accounts, by having
the *client* convert the utf8 input into koi8r, and then putting it
into the "octets" field. This of course requires that the user is able
to explicitly configure the client to assume that the server is
actually using koi8r.

Regards,
/Niels