Re: [Exim] Failing behviour based on SMTP codes.

On 2000-10-10 at 20:34 +0100, Philip Hazel gifted us with:
> Oh, wait. Are you saying you get a 550 upon connection? Or in response
> to the EHLO command? In those cases, yes, Exim treats it as a problem
> with the host, and will try the message again later. Maybe it shouldn't.
> Maybe that should be configurable, but at present it isn't.

When I read transports/smtp.c:smtp_deliver() it reads as though any
response at this stage starting with a '5', except in the case of EHLO
before HELO, causes an set_errno() call with fourth parameter 'FAIL',
in line 1147 (3.16 release), with the alternative being 'DEFER'.

What am I missing?

On 2000-10-10 at 16:44 -0400, Greg A. Woods gifted us with:
> Strictly speaking a 5xx error is a permanent error no matter where in
> the conversation it occurs. RFC-821 Appendix E strongly suggests a
> bounce is the only sensible action, even if the error can be corrected,
> since only the sender will know how best to handle the message.

This is also the assumed default behaviour behind RFC 1846
(Experimental) "SMTP 521 Reply Code" which is supposed to avoid any
extra coding logic. It documents that a bounce message should be
returned because of the 521 return code, and this whole RFC is
explicitly designed to prevent undeliverable mails waiting around in
SMTP queue purgatory for days, until things time out.
--
Civilisation: where they cut down the trees and name streets after them.