From: Andrew Gierth (andrew@erlenstar.demon.co.uk)
Date: Mon Jan 07 2002 - 02:51:51 CST
>>>>> "Claus" == Claus Färber <list-ietf-wg-apps-usefor@faerber.muc.de> writes:
Claus> It depends how you define the scheme. But yes, for reasonably
Claus> effiecient schemes, it would be completly unreadable;
Claus> something like se.test.+sdvjn23eor, WHICH IS STILL BETTER THAN
Claus> SEEING A NAME LIKE se.test.r||ksm||rg||s AND NOT BEING ABLE TO
Claus> USE THE GROUP.
If you see a group name se.test.+sdvjn23eor, how do you know whether
it is an encoding of a name with 8-bit characters or just a group that
happens to have a name component starting with '+'?
What do you do about those systems that currently refuse to
auto-create groups with such names?
Claus> It is debatable if the problem really existed, i.e. if mail
Claus> really was not 8bit clean
mail really was not 8bit clean, and some of us are to this day behind
mail relays that are actively 8bit-hostile.
>>> 6. A sysadmin can't create the newsgroup.[7]
>>> 7. A sysadmin creates the newsgroup incorrectly due to
>>> character set conversion issues.[7]
>> Which is likely to be a problem in a scheme where the logical name
>> of the group is se.test.räksmörgås and the physical name is something
>> else.
Claus> It is safe to assume that they can cut and paste and ASCII
Claus> sequence. It is not safe to assume that it will survive all
Claus> charset transformations.
It is not safe to assume that people are capable of cutting and
pasting accurately; misspellings of newsgroup names are common enough
even when they are sensible words and not nonsense strings.
Oh, and to take another of your points, NNTP is switching to UTF-8,
and there are no reports of any existing commonly-used news server
that doesn't already handle it sensibly (with the exception of the ?
metacharacter in wildmats for which support is patchy anyway).
-- Andrew.