ccfj@hippo.ru.ac.za (F.F. Jacot Guillarmod) (04/04/91)
The last thing this posting is intended to provoke is a religious war,
but I have this horrible feeling ;-) that it is going to do just that.
We are running two systems, one (a uucp <-> SMTP gateway) uses smail 3.1
and the other (an SMTP leaf node) runs sendmail. There has been a
bit of local confusion as to what is a 'correct' header, triggered off
by what seems to be a bug in elm 2.3 PL11. (see comp.mail.elm for a
description of how this 'bug' manifests itself).
The problem (if it is one) seems to be that the sendmail.cf file being
used on the leaf node is rewriting any address containing a '.uucp' in it
to a pure bang path format - e.g. user@node.uucp is being rewritten
to become node!user. In addition, it seems that the same sendmail rules
are prone to producing mixed form addresses such as c!b@some.domain.
The questions are:
a - is the behaviour of sendmail 'normal' if it does this?
To be more precise, on a system that is only connected via SMTP, are
mixed mode addresses and converting perfectly valid FQDN's to bang
paths 'OK'?
b - if this behaviour is not OK, does anyone have sample sendmail
rulesets that will allow a leaf node (i.e. not a gateway) that talks
only SMTP to conform to whatever standard is deemed appropriate?
The present sendmail.cf is based very closely on the SunOs release
version in /usr/lib/sendmail.subsidiary.cf. The behaviour described
here seems due to the way ruleset S4 treats '.uucp' addresses.
c - is there some reason that sendmail rulesets behave like this? If,
as I suspect, it is due to historical reasons, shouldn't there be
some other default .cf files doing the rounds at this stage of the
networking game? A large number of systems out in the wide world
seem to behave in this way (a good source of data is having a look
at the vote acknowledgement postings in news.newsgroups) but does
this make them 'right'?
The following sample headers were received from the same sender, but on
our two different systems. To my inexperienced eye, it seems that the
first set is internally consistent and conforms strictly to RFC822
standards. The second set of headers, to my other inexperienced eye,
look 'funny'.... but are they 'wrong'?
Mail headers after incoming mail is processed by smail 3.1.18:
=======================================================================
From proxima.uucp!frcs.uucp!paul Mon Apr 1 13:54:53 1991
Return-Path: <@proxima.uucp:paul@frcs.uucp>
From: paul@frcs.uucp (Paul Nash)
Subject: Does mail from proxima arrive there?
To: ccfj@quagga
Date: Mon, 1 Apr 91 13:13:31 SAT
========================================================================
Mail headers after being processed by sendmail:
=========================================================================
From proxima!frcs!paul@Daisy.EE.UND.AC.ZA Tue Apr 2 12:40:22 1991
Return-Path: <proxima!frcs!paul@Daisy.EE.UND.AC.ZA>
From: frcs!paul (Paul Nash)
Subject: Re: Mail paths
To: ccfj@hippo.ru.ac.za (F.F. Jacot Guillarmod)
Date: Thu, 28 Mar 91 8:33:58 SAT
==========================================================================
(please note that the delivery path to these hosts is slightly different)
This isn't really an argument about the merits of smail vs sendmail - it
is apparent that sendmail can be made to behave in the appropriate way,
if only an appropriate way can be defined and then somehow converted to
a set of rules. Any assistance in this regard would be appreciated.
--
F.F. Jacot Guillarmod - Computing Centre - Rhodes University
Artillery Road - P.O Box 94 - Grahamstown - 6140 - South Africa
Internet: ccfj@hippo.ru.ac.za Phone: +27 [0]461 22023 xt 284
uucp: ..!uunet!m2xenix!quagga!hippo!ccfj Fax: +27 [0]461 25049jsd@proxima.UUCP (Jeremy Gryttpype Druker) (04/05/91)
ccfj@hippo.ru.ac.za (F.F. Jacot Guillarmod) wondered about sendmail rewriting user@host addresses as host!user, or host!user@host... I think I recall similar behaviour on proxima at one stage, but it was around that time that we replaced sendmail with smail ;-) It's a bit hazy, but I think I figured out that sendmail did indeed have "prior knowledge" of uucp - over and above the sendmail.cf file - and rewrote addresses accordingly. There may me a mailer flag involved in this, but I've just checked the FM and nothing rings any bells. If the thought of trudging through the sendmail.cf file isn't utterly appealing, there is a package called Ease that appeared recently in alt.sources, which claims to be able to translate a sendmail.cf file into human-readable form, and then back again. Mail me if you don't have it there. -J ---------------------------------------------------------------------------- In a distant and second-hand set of | Jeremy Druker dimensions; on an astral plane that | jsd@proxima.UUCP was never meant to fly... | ...!uunet!ddsw1!olsa99!proxima!jsd --------------- - Terry Pratchett ----------------------------------------
ccfj@hippo.ru.ac.za (F.F. Jacot Guillarmod) (04/05/91)
In <4594@proxima.UUCP> jsd@proxima.UUCP (Jeremy Gryttpype Druker) writes: >ccfj@hippo.ru.ac.za (F.F. Jacot Guillarmod) wondered about sendmail >rewriting user@host addresses as host!user, or host!user@host... >If the thought of trudging through the sendmail.cf file isn't utterly >appealing, there is a package called Ease that appeared recently in >alt.sources, which claims to be able to translate a sendmail.cf file >into human-readable form, and then back again. Mail me if you don't >have it there. I think the point here is to find out what sort of consensus there is (if at all possible) on what constitutes 'acceptable' headers, and only then to try and formulate a set of rules (if at all possible) to implement this behaviour. The default distribution sendmail.cf's were obviously formulated the way they are for some sort of reasons, and it may be instructive to find out what those reasons were, and if they are still valid. A large number of sites _must_ be using configurations close to the default releases - and finding them satisfactory. Is internet purity or rfc822 compliance a myth? Thanks for the tip, and offer..... -- F.F. Jacot Guillarmod - Computing Centre - Rhodes University Artillery Road - P.O Box 94 - Grahamstown - 6140 - South Africa Internet: ccfj@hippo.ru.ac.za Phone: +27 [0]461 22023 xt 284 uucp: ..!uunet!m2xenix!quagga!hippo!ccfj Fax: +27 [0]461 25049