On Sun, 21 Mar 1999, Leonardo Boselli wrote:
> I had to put a machine listening on that address and beside
> forwarding to the correct address, bouncing back the message
[...]
> From: "john Smith" <js@???> (By way of lazy@??? )
> Of course the autoresponder replied to original sender.
> What would have been the correct way to operate, both in out and
> their part ?
By "on their part" do you mean lazy@???? It's not his fault, except
for using Eudora, which puts that silly comment in the From: field
instead of properly inserting a Resent-From:.
Bounce messages and other error reports should be sent to the SMTP
reverse-path, that is the address in the MAIL FROM: element of the
protocol conversation, which frequently (but not always, though it's a
server bug when it does not) appears in the Return-Path: field. (See
RFC821 sections 2, 3.1, 3.2, 3.6, 4.1.1, RFC1123 sections 5.2.8, 5.2.13,
5.3.3, and RFC822 section 4.3.1, 4.4.3, 4.4.4). Note that this means that
the Return-Path: should have been set to <lazy@???> on delivery of the
re-sent message, replacing the Return-Path: <js@???> field if one
was present.
If the Return-Path: was incorrect, the SMTP server on the machine where
you set up the .forward file needs repair. If your autoresponder used the
From: address instead of Return-Path:, then either the autoresponder needs
repair, or you used an autoresponder which conforms to RFC822 4.4.4, in
which case it's not an approriate responder for generating error reports.
In general, errors of this sort should be mailed back by the SMTP server
itself rather than by a program under user control (like .forward). One
way to cause this to happen without having to reconfigure your SMTP server
is to have the program that is run from .forward print a suitable response
to its standard error stream, and then exit with a nonzero status. The
SMTP server will interpret this as failure of the .forward program, and
(in most common configurations) will therefore send back a bounce message
containing the error output.
--
*** Exim information can be found at
http://www.exim.org/ ***
