Re: smtp talking to loopback

[Matti Aarnio <mea@nic.funet.fi>]

> I have found that in smtp.c there is code to prevent me
> from sending mail back to myself. Why is this considered
> bad??

	A last-line defence against MX pointer failure.
	It is troublesome with explicite routes / when you
	want to test weird things / ...

>   What I want to do is send mail back to myself on
> a different port (where I have a sendmail running) to 
> perform some special processing. Maybe there should be
> a check for a port other than the SMTP port, and if
> that is the case, then figure that the maintainer knows
> what they are doing and let it through. Does this
> seem reasonable???

	Yes.  If the destination port != IPPORT_SMTP (25),
	then it is ok to talk to myself..  (unless the same
	ZMailer is listening at that weird port, and ...)
	(I am happy to declare that if sysadmins are doing
	 that sort of tricks, they are all on their own, and
	 ZMailer is better to let them just do it.)

	I have added this to my 2.99.48p3+ code base too.


	I will dump  2.99.49b11 (without 'aa') to my workstation
	anonymous ftp before I head for home today.

	It contains also partially coded policytest facilities for
	the smtpserver.  It is essentially Gabor Kissigs algorithm,
	but done entirely in the smtpserver.
	(But it does not work yet! See comments at the source.)

	Oh yes,  doc/design/zmog.ps  is more up to date, than the
	old info file, but still not completely up to date..
	We will be using that text in the new documentation too.

> /mrg

	/Matti Aarnio <mea@nic.funet.fi>