On Mon, Mar 17, 2003 at 11:43:07PM +0000, Giuliano Gavazzi wrote:
> At 22:13 +0000 2003/03/17, Matthew Byng-Maddick wrote:
> >You will also note, then, that the 5xx on connect or after the HELO has
> >been illegal (other than the illegal argument) since about that time. This
> >is for precisely this reason. (Remember: you MUST accept a postmaster@
> >address).
> but you know that "554 on connect" has been indeed introduced in RFC2821.
Yes, I think it's ill-advised.
> >If I can't send mail to postmaster, then your MX is breaking the RFCs, and
> >therefore you can't expect RFC compliance from the other end.
> RFC2821 states that postmaster is not strictly necessary if you send
> 554 at all connections openings... Understandable.
I know. I still think it's ill-advised.
> >If you think it's a sane thing to do, think again. It is stupid to do a
> >rejection before RCPT under any circumstances.
> I agree, nevertheless the RFC seems to give some consideration to
> this 554_on_connect, I was only proposing a solution to a problem (if
> you like "an interpretation").
Indeed.
> My proposal was not to treat all 5xx as a 4xx, but was to treat 554
> on connection effectively as 421 on connection + add a warning to the
> postmaster (for configuration debugging). It was based on three
> keywords: MX,timely delivery, and 554, all missing from RFC821. It is
> time to leave 821 for 2821.
With issues like this, I'm not so completely sure. I think that all in all
821 is a better standard. I think that returning 554 as a greeting on a
public-facing MX is probably pretty stupid. Giving an RST is so much better.
MBM
--
Matthew Byng-Maddick <mbm@???> http://colondot.net/
