Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!seismo!lll-lcc!ames!ucbcad!ucbvax!ucbarpa.Berkeley.EDU!fair
From: fair@ucbarpa.Berkeley.EDU.UUCP
Newsgroups: news.sysadmin,news.admin
Subject: Re: Negative Article ID's
Message-ID: <18662@ucbvax.BERKELEY.EDU>
Date: Fri, 1-May-87 06:25:09 EDT
Article-I.D.: ucbvax.18662
Posted: Fri May  1 06:25:09 1987
Date-Received: Sat, 2-May-87 14:15:28 EDT
References: <3673@drutx.ATT.COM> <1987Apr27.131502.8062@sq.uucp>
Sender: usenet@ucbvax.BERKELEY.EDU
Organization: USENET Protocol Police, Western Gateway Division
Lines: 37
Xref: utgpu news.sysadmin:167 news.admin:331
Summary: message-ids are NOT just "unique strings"

In article <1987Apr27.131502.8062@sq.uucp> msb@sq.UUCP (Mark Brader) writes:

	It should perhaps be pointed out, in case anyone else is
	confused, that the Message-ID in news is simply a string, and
	any site is allowed to generate it in any format they like.

This is absolutely, flatly FALSE. There is a standard for USENET
message headers, defined in the document RFC850, which is included in
the B news 2.11 distribution. It cites that the message-id should be in
RFC822 syntax, excepting that a USENET message-id should never contain
linear white space (spaces or tabs). RFC822 is even more grindingly
specific about the format and content of a message-id; I once wrote a
parser/syntax checker for it.

	For instance, check out the Message-ID on THIS article.  It was
	generated by C news (the as-yet-unreleased system described by
	Geoff Collyer and Henry Spencer at the last USENIX conference).
	I presume they use this format because it is easy to generate
	reliably in sh.

It does happen that the message-id that you cite (reproduced in the
first line of this article) is legal according to the specifications.
Knowing both Henry & Geoff as I do, I really doubt that it was an
accident that the format is legal (and, as you note, trivially
generated from a shell script).

When I get around to recoding the USENET to ARPANET side of the gateway
on ucbvax (the process in that direction is mostly an AWK script, and
it will be handling increasing volumes as time goes on, so it is now
time to recode it in C; going the other way, it's already in C: a new
version of recnews), the message-id syntax checker will be included in
that program, and it will refuse to gateway any USENET article that
does not conform to spec.

	where is Mr. Protocol when I need him?

	Erik E. Fair	ucbvax!fair	fair@ucbarpa.berkeley.edu