> Date: Sat, 31 Jan 2004 15:20:27 +0000
> From: David Woodhouse <dwmw2@???>
> To: Ian A B Eiloart <iane@???>
> CC: Exim users list <exim-users@???>
> Subject: Re: [Exim] Temporary defer on callouts
>
> On Sat, 2004-01-31 at 13:09 +0000, Ian A B Eiloart wrote:
> > I'm not sure whether the RFC permits this, though. As I understand it,
> you > must accept mail to postmaster, and to any known mailbox (but can
> reject > mail addressed to a non-existant mailbox).
>
Just to clarify:
This is completely different than 'blackholing' or just plain not
delivering mail from a null user at a later time.
These are two seperate issues and callout from exim will fail this test
to att.com mail servers.
220 att.com - Maillennium ESMTP/MULTIBOX kcmsi #175
helo domain.com
250 att.com
mail from: <>
552 null MAIL FROM not permitted
And for those who have not figured it out yet, this does mean that you
can not send 'reject/bounce' messages to att.com users, so they never know
if there is a delivery problem after it leaves the att network.
>
> Although RFC2821 says you MUST send bounces with empty reverse-path, it
> doesn't actually say you MUST accept them. You might be expected to
> infer that for yourself in the general case though, given a modicum of
> common sense.
>
> OTOH in the absence of specific instructions to the contrary, it does
> seem perfectly valid to refuse to accept mail with null reverse-path
> even for known-valid addresses, _if_ it's known that those addresses
> never send mail, and hence should never receive bounces.
>
> Refusing to accept bounces to mailing lists, for example, does make a
> lot of sense. Likewise for old email addresses which are forwarded to
> current addresses only for backward-compatibility.
--
--EAL--
