From: Kai Henningsen (kaih@khms.westfalen.de)
Date: Tue Nov 04 1997 - 14:57:00 CST
chl@clw.cs.man.ac.uk (Charles Lindsey) wrote on 04.11.97 in <9711040958.AA00346@clw.cs.man.ac.uk>:
> sommar@algonet.se (Erland Sommarskog) wrote:
>
> >Charles Lindsey (chl@clw.cs.man.ac.uk) writes:
> >>No, you can't do that, because a line in the newsgroups file is already
> >>expected to fit within 79 characters.
>
> >Stupid question: is the 79-char limit defined in an RFC somewhere
> >or is just convention? I have to say that it seems too narrow anyway.
> >And with newsgroups file not being used by many readers, I'm skeptic
> >to have size limits because of it.
>
> Just because some broken readers don't give you access to the newsgroups
> file ...
>
> Personally, I find myself referring to that file frequently.
>
> However, we are trying, elsewhere in our document, so say that lines
> intended to be read by humans (notably the body) SHOULD fit into 79
> characters (that is why we are trying to allow Newsgroup and Path headers
> to be wrapped). Lines in the newsgroups file are also intended to be read
> by humans. Ergo ...
Well, the newsgroups file format has the obvious problem that it doesn't
have enough room if you limit to 79 characters. Such a limitation is
annoying, for the same reason that articles written with 40 character
lines are annoying. (That's what a 79 character limit for the newsgroups
file essentially amounts to.)
Present that file with two lines per group, one for the name, one for the
description. No problem at all.
If hierarchies want to adopt such a limitation, that's their problem. I
see no reason to write it into the standard.
> And in trying to manage the uk.* hierarchy, I have to expend much effort
> in persuading writers of RFDs to provide conforming newsgroup lines. The
> only definitive text on the matter at the moment is a FAQ by Russ. Perhaps
> he would care to comment. (BTW Russ, when is that FAQ going to have a
> permanent archive with a URL?) I think at some points in our document we
> are going to have to specify limits that are for human convenience rather
> than technical necessity. After all, humans are the ultimate customers. We
> are not just trying to conctruct a means for computers to speak to one
> another for their own amusement :-)
In that case, I'd want the spec to say that descriptions should have some
_minimal_ length - say, at least 40 characters :-)
> At any rate, for human convenience, I think the length of newsgroup
> components should be fixed at one of 14, 16, 24 or 32 (visible characters,
> that is).
You have a truely weird definition of human convenience. I don't share it.
*Convenient* is if I can make components, and newsgroup names in general,
as long as I want to.
> You will break existing software (notably CNEWS) if you permit characters
> other than lowercase, '+', '-', '_' in newsgroup names. What is the
*Break* Cnews? I'd like to see proof for that assertion. Especially as
I've seen active files that contained such groups.
> benefit? To allow people to create even more joke names in alt.*?
alt is lost anyway. There's no way I can see to save that hierarchy. Two
or three years ago, it might have been possible, but no longer.
As for the rest of the world, if there's not a really good reason to
forbid one of these characters, it ought not to be done.
In German, for example, the lowercase rule means that _all_ nouns in
newsgroup names must be written wrong (because in German, all nouns start
with an uppercase letter). Now this may or may not be a problem, but I see
no good reason for the standard to define the one right way here.
MfG Kai