From: Bruce Lilly (blilly@erols.com)
Date: Sun Jun 08 2003 - 17:32:07 CDT
Charles Lindsey wrote:
> In <3ECF9CB6.3040204@Sonietta.blilly.com> Bruce Lilly <blilly@erols.com> writes:
>
>
>>... and as newsgroup names appear in header fields, if length of
>>those names (and/or their components) is an issue at all, we should
>>say something relevant in USEFOR. As discussed earlier, obviously the
>>common message format maximum line length (in the absence of a documented
>>mechanism for splitting and combining names, such as permitting line
>>folding between components) is a technical limit on length. We probably
>>don't need to mention that, as it's implicit in referring to 2822[*].
>>But there may be lower limits imposed by NNTP and IMAP protocols and/or
>>implementations, and if so we should incorporate those limits into the
>>syntax (with mention of the technical limit(s) as an issue in addition
>>to display where newsgroup name/component length is discussed in USEAGE).
>
>
> There are no real technical limits on the length of a newsgroup-name,
> though there are lots of social limits. NNTP has a 512 character limit on
> the length of commands (which sometimes incorporate newsgroup-names), but
> I don't think anybody would seriously contemplate newsgroup-names that
> long.
I specifically mentioned NNTP. In what way is that a social rather
than technical limit?
>>Note:
>>* unless the must-have-non-whitespace-content-on-first-field-line rule
>>remains a MUST, in which case we need to at least note that a long
>>newsgroup name that fits on the first line of a Newsgroups header field
>>is still unacceptably ling because it will fail to fit on the first
>>line of a Followup-To header field.
>
>
> You have just produced a good reason to limit the length of a
> newsgroup-name to 66 characters in place of the 71 in our present draft.
No, I've produced yet another example of why the must-have-non-whitespace-on-
first-field-line rule is a problem.
> Would anybody object if I changed that 71 to 66? Note that it is, in any
> case, only an advisory limit which hierarchy admins can choose to follow
> or not (and will be discussed in more detail, in USEAGE).
Don't blame it on me. I proposed removing the magic numbers (unless there
is sound technical basis for a limit (e.g. NNTP commands, dbz) in favor
of simply presenting the issues. IIRC Martin Duerst indicated that he felt
that was the right direction. Removing the magic numbers also obviates
the issue of whether the numbers refer to characters, glyphs, "columns",
etc, which seems to have been the only other issue discussed (and not yet
resolved).
>>At one time, 79 might have been an appropriate magic number, but
>>that time (if it in fact existed) has passed.
>
>
> Why do you suppose that 79 is no longer the appropriate magic number?
For reasons presented in detail previously. Quo vide.
> Anyone who posts articles with lines longer than that without good cause
Are we still talking about header fields, or have you changed the topic
to length of body text lines?
And no, I didn't produce a reason to use an even lower limit because
some followup agents put attribution lines with text longer than
"followup-to: " in article bodies. That's not our issue.
> (long URLs are the usual counter-example) is likely to be flamed in most
> places on Usenet, as is anyone who complains that his reader will not
> accept lines of at least that length.
If you're going to put in specific number limits for other than technical
issues, then you might as well prohibit newsgroup components of length 13
because some people suffer from triskaidekaphobia, and of length 4 for similar
reasons (one of the Japanese words for 4 is pronounced the same as the word
for death, and so many Japanese people are superstitious about that number).
And while you're at it, prohibit numeric components with those values also.
And don't forget to deal with the other "magical" numbers, 7, 666, ...
For that matter, you might as well mandate that UAs change font size (if
they are capable of doing so) so as to display newsgroup names on one display
line where ever those names appear (in header fields, in attribution lines,
anywhere in a message body, ...). Don't forget to explain exactly how a
UA is supposed to recognize a newsgroup name in the middle of body text...
All of which is neither more nor less silly than mandating a limit based on
the number of columns on a teletype printer (a device which has long been
obsolete).