From: Kai Henningsen (kaih@khms.westfalen.de)
Date: Sun Nov 02 1997 - 07:09:00 CST
chl@clw.cs.man.ac.uk (Charles Lindsey) wrote on 14.10.97 in <9710141928.AA13584@clw.cs.man.ac.uk>:
> NOTE: The 14 character limit on the length of a
> component arose originally from a limitation in the
> length of a filename in early UNIX systems. Although
> that is hardly a reason to retain that particular limit
> at the present time it is useful, for reasons of human
> comprehensibility, to have some limit. Since, then, an
> arbitrary choice had to be made, it seemed simplest to
> leave it at 14, as being in the general region from
> which any choice would reasonably have been made.
> [Only a flamefest could agree on any other number]
I think this is unacceptable.
However, I think it would be acceptable to say that
newsgroup names MUST be unique if every component is shortened to 14
octets (independent of legality in UTF-8)
OTOH, I don't think there really is any technical justification left even
for this rule. Limit total newsgroup name length to 255, and leave it at
that, would be my preferred solution.
As for allowed characters, it might make sense to forbid all-numeric
components, as those make problems for a lot of existing implementations
(like C-News and INN), plus all those fixed strings that have been used
differently. Anything else should only be cases of obvious potential for
harm, which essentially reduces to some specific ASCII characters like
":%/&*?" (exact set TBD, but think "URL syntax" and "special shell
characters". No non-ASCII characters should be on that list, and certainly
not ASCII upper case. Leave that to hierarchy-specific rules.
MfG Kai