[Greylist-users] Re: GroupWise is the Devil

[Scott Nelson]

At 10:10 PM 7/6/03 -0500, you wrote:
>Scott Nelson <scott at spamwolf.com> wrote:
>>Note that even if this bug is fixed, there will still be a lot
>>of broken mailers for a long time.
>
>It's a given there's a few really broken ones, hopefully only a very few
>like Groupwise.
>
>
>>The solution is to fail the DATA with a 4xx /as well/ if there
>>was at least one RCPT that was 4xx.
>
>If you check out the URLs marius posted, you'll see the Groupwise SMTP
>exchange with the destination system looked like this:
>
>  220 megatron.alpha1.net. ESMTP Sendmail 8.1
>  EHLO apcis7.tamu.edu.
>  250-megatron.alpha1.net. Hello cis-gw.tamu.
>  MAIL FROM: <philip-kizer at cis-gw.tamu.edu>
>  250 2.1.0 <philip-kizer at cis-gw.tamu.edu>
>  RCPT TO: <marius at alpha1.net>
>  451 4.7.1 Please try again later (TEMPFAIL
>  Quit
>  221 2.0.0 megatron.alpha1.net closing conn
>
>That did not even leave an option of responding to the DATA phase with a
>4xx since it didn't begin one.  Groupwise is just broken, such as per
>RFC2821, notably sections 4.2.1 and 4.5.4.1 ("mail that cannot be
>transmitted immediately MUST be queued and ... retried", as opposed their
>taking a 451+4.7.1 and immediately turning it into a message claiming there
>was a "551 No valid recipients" response).
>

I don't think it's reasonable to expect programs written prior
to April 2001 to conform to RFC2821.  RFC821 only says
"the sender-SMTP is encouraged to try again."
But that's no excuse for turning a 451 after RCPT into a 551.

I'm not claiming it's not broken, but my experience with these
kinds of problems tells me that it won't be fixed in anything
approximating a timely manor.