On Mon, 11 Mar 2002, Dave C. wrote:
> It would be nice to have a a way to say 'for these senders' or 'for
> these hosts', issue 5xx codes for any and all received commands.
Well, you could only do it 'for these senders' after a MAIL command had
been received. You couldn't, for example, reject HELO 'for these
senders'.
And I don't think it would be right to reject RSET 'for these senders'
either.
In practice, if I were to implement an ACL for MAIL, you could make it
issue a 5xx command. Then any subsequent RCPT or DATA commands would
automatically be rejected because a valid MAIL is needed first.
> It would be a nice option for some future version. Im currently dealing
> with some MTA that seems to be ignoring my rejections of his RCPT...
> sigh..
How would rejecting MAIL help? If the MTA is that broken, it probably
ignores rejections of MAIL as well.
Question: How do you know it is ignoring your rejections? There is a
possibility for confusion here if PIPELINING is in use. When it is, the
client is entitled to send
MAIL...
RCPT...
RCPT...
...
DATA
all in one packet, and then pick through the responses that the server
sends. So, in this scenario, if you reject all the RCPTs, you'll still
see the DATA (which you reject, of course). But the client is not
behaving improperly. However, if it subsequently tries to resend the
message, it is certainly broken.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.