Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site nsc.uucp
Path: utzoo!linus!allegra!eagle!harpo!seismo!hao!menlo70!nsc!glenn
From: glenn@nsc.UUCP
Newsgroups: net.news.b
Subject: header.c (r2.19) strangeness
Message-ID: <306@nsc.uucp>
Date: Fri, 3-Jun-83 22:18:26 EDT
Article-I.D.: nsc.306
Posted: Fri Jun  3 22:18:26 1983
Date-Received: Mon, 6-Jun-83 16:48:35 EDT
Organization: National Semiconductor, Sunnyvale
Lines: 33

I've installed the changes to header.c that Mark sent out a few days ago
(bringing it to version 2.19) and have found what seems to be a bug.

One of the changes from the 2.10 distribution version is to add the variable
seenrelay, which indicates whether a "Relay-version:" header line has been
seen yet.  The variable is used in the routine type() to determine whether a
"From:" header line is to be treated as a FROM line or a PATH line.  The change
is:
	if (its("From: "))
!		if (index(ptr, '@') && !index(ptr, '!'))
			return FROM;
		else
			return PATH;
***********
	if (its("From: "))
!		if (index(ptr, '@') && !index(ptr, '!') && !seenrelay)
			return FROM;
		else
			return PATH;

I don't understand why this variable is relevant to the decision.
Including it makes readnews think that legitimate "From:" lines aren't,
which in turn causes article headers (in the msgs interface, at least),
to be less informative than they should be;  instead of generating
"machine!person (Real Name @ Organization)", readnews generates only
"machine!person".

Can someone explain why seenrelay exists and whether this effect is
intentional?

		-- Glenn Skinner, National Semiconductor
		...!{fortune,menlo70}!nsc!glenn
		(408) 733-2600 x 335