Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
Subject: Re: Incessant Bounces
From: Greg A. Woods (woodsweird.com)
Date: Thu Aug 24 2000 - 11:41:04 CDT

[ On Thursday, August 24, 2000 at 09:08:47 (+0200), Brad Knowles wrote: ]
> Subject: Re: Incessant Bounces
> You're welcome to set up your mail systems any way you like.
> However, you are fundamentally unable to provide an argument that the
> default method of handling response codes in these uncertain cases is
> to do anything other than issue a 4xx and allow the sender to retry
> at a later time.

Ah, Brad, let's get something straight here. If I correctly interpret
your analysis of the events at AOL which caused you to fight this fight,
there's absolutely no difference in the result between what actually
happened and any other kind of programming or operational error that
could have been made in the AOL mailers or nameservers. I.e. your use
of this event to claim that mailers should not by default bounce e-mail
when they get an authoritative NXDOMAIN reply is totally bogus. The
event was entirely caused within AOL itself and the same result would
have occurred even if the cause was a programming error in AOL's
mailer. Claiming that AOL's error can happen to anyone is meaningless.
Yes, people do make errors in programming and errors in operating
nameservers, etc., but those are real errors and they should cause real
bounces of e-mail because that's one of the best ways to bring the error
to the attention of those who it effects.

So, from what I've been able to gather you are in effect using only my
own evidence of invalid authoritative NXDOMAIN as the sole reason for
justifying your cause (remember I told you about this claim of mine
several times in the past).

If that is not true then please provide your own evidence of current or
past invalid authoritative NXDOMAIN replies. I withdraw my claim
because I have not experienced any similar problem in the past 3-4 years
or more and I can no longer even produce documentation showing the
original problem. It may well have been a mistake on my part -- I've
since learned even more than I knew then about diagnosing strange
problems in the DNS. Maybe what I saw was indeed a valid NXDOMAIN.

In the mean time even if there are mistakes in 0.00001% of DNS queries,
I doubt anyone really wants *all* of their undeliverable e-mail to be
held in limbo for hours, days, or weeks, just to prevent a tiny fraction
of invalid bounces from happening! I doubt people even want a fraction
of their valid bounces being held in limbo. Delaying a delivery or a
bounce unnecessarily is wrong.

Telling me that I have to provide mathematical proof that some kind of
error can never ever happen, particularly when we both know that such a
proof is impossible, is mis-directing away from the real issue here.

							Greg A. Woods

+1 416 218-0098 VE3TCP <gwoodsacm.org> <robohack!woods> Planix, Inc. <woodsplanix.com>; Secrets of the Weird <woodsweird.com>