From: Erland Sommarskog (sommar@algonet.se)
Date: Wed Nov 19 1997 - 17:09:51 CST
Leonid Yegoshin <egoshin@genesyslab.com> writes:
>To: Chris Newman <Chris.Newman@innosoft.com>,
> =?iso-8859-1?Q?Martin_J=2E_D=FCrst?= <mduerst@ifi.unizh.ch>
>Cc: Erland Sommarskog <sommar@algonet.se>, usenet-format@clari.net
Please, there's no reason for this massive cc:ing. We are all on
the mailing list, aren't we?
> The idea of using UTF-8 has yet one bad consequence - all draft use MIME
>only and I can implement MIME-only software without problems. But if we
>insert possibility for UTF-8 for 8-bit group names we should have
>an additional task with processing of UTF-8 in news software.
>
> Moreover, it can be another problem if we adopt UTF-8 default:
>if some article pass UTF-8 host then newsgroup names CAN be converted
>from MIME to UTF-8 by some energetic news manager !
>(Especaily if he has UTF-8 compatible local host)
I'm not sure that I understand what you are saying here, but I like
to point out that our previous discussions have led to the conclusion
that the Newsgroup header more or less must have a fixed charset,
and that UTF-8 is the only reasonable choice. The newsgroup name
must be passed untouched from server to server, and from server to
client, with the only exception that client will need to convert
forth and back to the client charset.
Any attempt to include a coding à la RFC2047 is likely to lead to
names that are completely unbearable for anyone who is seeing the
name with a non-MIME-capable tool - and that includes the sysadmin
who might be grepping his active file. (Or any regular user who
just telnets to port 119 and say LIST ACTIVE, as I do every once
in a while.)
-- Erland Sommarskog, Stockholm, sommar@algonet.se