Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version VT1.00C 11/1/84; site vortex.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!gamma!epsilon!zeta!sabre!petrus!bellcore!vortex!lauren
From: lauren@vortex.UUCP (Lauren Weinstein)
Newsgroups: net.mail.headers
Subject: Re: Mail Header Summary Document...
Message-ID: <841@vortex.UUCP>
Date: Wed, 16-Oct-85 14:13:47 EDT
Article-I.D.: vortex.841
Posted: Wed Oct 16 14:13:47 1985
Date-Received: Fri, 18-Oct-85 00:31:58 EDT
References: <50000001@hpcnof.UUCP>
Organization: Vortex Technology, Los Angeles
Lines: 56

I do hope people aren't going to take that document in its current
form and use it as a reference.  In particular, the descriptions of

From_ (From followed by space)
>From_
From:
Return-Path:
Sender:

are woefully inadequate and misleading.  By failing to indicate the
important variations between uucp format and 822 format use of such
headers, all sorts of confusion may result.  For example, saying that
smart sites SHOULD update the From: line is very misleading.  Updating
of that sort has caused lots of problems since the From: line (which
is supposed to be an 822 format line) is not subject to uniform
handling across the networks.  Especially when @-sign addresses are
present, but in general even when only ! addresses are present, the
From:, To:, and other 822 lines should be LEFT ALONE by all intermediate
sites.  The uucp From_ line (and >From_ lines), must
either be left on messages to provide routing info or else folded
into a single From_ line (NOT a From: line unless the site doing the
folding is absolutely certain it is folding all info from the various From_
and >From_ lines CORRECTLY, and only when @-addresses are not in use).

If a site insists on folding From_ and >From_ lines, it should never do
so unless Received: lines are being added, since otherwise important
date/time transit info is lost.  If a site is going to try create
a new From: line from the From_ and >From_ lines, it must do so only
completely on that basis, NOT by simply trying to update the existing
From: line which may rooted off of a different address.  If an @ address
is present it must not try do this under any conditions, since incorrect
hybrid addresses almost always result.  And the site
must still always be sure to properly add a new From_ line, update the >From_
lines or create a new folded From_ line as appropriate.  The existence
of From_ and >From_ lines is critical.  Once again, the best solution
is to never count on the From: line as being an accurate route back
to the author unless it is in 822 @-form.  Such addresses, especially,
should not be touched by intermediate sites under any conditions.
Return-Path is also not reliable as a return address.  In general,
for pure uucp mail, the only valid return address for replies is that
generated from the From_ and >From_ lines.  Use of the From: line
is possible in some cases, especially when @-addresses are present
instead of ! addresses with smart mailers.  But if intermediate
sites insist on trying to do simple updates of the From: lines based
on existing From: lines, the information in the From: line frequently
will be incomplete.  The recommendation is that sites stop trying to update
the From: lines at all, both in the presence of @ addresses and in the
presence of simple ! addresses or hybrids.  Leaving From: alone in all
cases is probably the best solution in all cases in the long run.

As we move toward domain addressing, many of these problems with
From:, etc. will gradually become of less significance (assuming
we can get sites to stop modifying these lines and just leave them
alone in most cases!)

--Lauren--