Re: reject_multi_recipient_bounce & rejected recipients

Michael Tokarev:
> Currently, reject_multi_recipient_bounce checks whenever
> we are about to *accept* more than one recipient, and
> does not count other, rejected recipients, so it is
> possible to have session like this:
>
> MAIL FROM:<>
> 250 Ok
> RCPT TO:<unknown1@example.com>
> 550 Address unknown
> RCPT TO:<unknown2@example.com>
> 550 Address unknown
> RCPT TO:<known1@example.com>
> 554 Client host rejected: can not find your hostname, [127.0.0.1]
> RCPT TO:<known2@example.com>
> 250 Ok
> RCPT TO:<known3@example.com>
> 550 <known3@example.com>: Recipient address rejected: Multi-recipient bounce
>
> I think the reject_multi_recipient_bounce should trigger even
> at known1@example.com stage, and at least it should not permit
> known2@example.com. Technically it is easy to implement, by
> changing the test from evaluation number of accepted rcpts
> to using additional bit in state field which will be set when
> smtpd sees at least one rcpt. But the question first: is such
> behaviour ok?

What do you mean, are you proposing that Postfix no longer stops
at the first access restriction that results in a yes/no decision?

Wietse