At 05:42 AM 2002-10-26 -0700, Jeffrey Goldberg wrote:
>On Sat, 26 Oct 2002, Nick Simicich wrote:
>
> > >Then why send filter-bounces at all? Why can't the filtering or
> > >user-initiated blocking happen quietly?
> >
> > If it were a privately composed piece of mail that you sent to someone, you
> > might well want to know whether or not it actually got through. It is right
> > to return a bounce.
>
>I think the question is whether this should be considered a bounce (and so
>go to the envelope sender) or be considered an autoresponse (and so go to
>the header reply-to or header from).
It is never right to send any autoresponse regarding the message delivery
status to anything other than the RFC821 sender (or the contents of the
Sender: header, which should be specified if the RFC821 sender does not
match the From: line). No doubt this is heresy, but, for example, vacation
responses should always go to the sender and not to any of the other
entities mentioned in the RFC822 headers. The only time that the RFC822
headers should be used for generating responses is when there is human
interaction. This is especially true for responses generated by MTAs.
RFC822 4.4.4 attempts to discuss this, and, to some extent, my feeling is
that Return-Path has supplanted sender for this usage, and when looking at
how smtp mail is delivered, the RFC821 MAIL FROM: header is moved into the
sender (or Return-Path) when the mail is "gatewayed" from a RFC821 context
to an RFC822 context, that is, when it is delivered. At the very least,
any notification of the status of the message should be sent to the
sender/Mail From: header and only responses to the message should use the
RFC822 From: or Reply-to: headers.
This is clearly a notification of non-delivery, and not a response.
> > The real issue is whether it is right to bounce mailing
> > list mail.
>
>As you say below, it is very hard to distinguish. I have a rant about
>what autoresponders should and shouldn't do. But it isn't clear that this
>is should be an autoresponder.
Again, I have no idea where you got the idea that autoresponders that are
reporting delivery status should send to anything other than the RFC821
header or, as some MTA/MUA combinations (depending on which side of the
line you put a delivery agent) the Return-Path: header. If they never did
anything other than that, then vacation programs, for example, would never
bother anyone other than the list (they would never bother people who
composed mail and sent it to the list).
For most cases, the addresses in the RFC821 and 822 headers should be the
same. For those cases where they are not, such as mailing list mail, I
can't think of one where it is not right to use the 821 headers.
> > The more I consider this, the more I feel that it is not right for AOL to
> > filter this mail at all based on the fact that the user instructed AOL to
> > block mail from an individual, but then mail came from a mailing list was
> > blocked instead. I think that this is a bug and someone who cares and who
> > has an AOL account should report this.
>
>I'm coming to agree with you (and am revising my initial stance based on
>what you and others have written). This really is an automated user end
>mail filtering. It shouldn't really be generating bounces.
>
>-j
>
>--
>Jeffrey Goldberg http://www.goldmark.org/jeff/
>Relativism is the triumph of authority over truth, convention over justice
--
Take The Boulder Pledge Today
"Under no circumstances will I ever purchase anything offered to me as the
result of an unsolicited e-mail message. Nor will I forward chain letters,
petitions, mass mailings, or virus warnings to large numbers of others.
This is my contribution to the survival of the online community." - Roger
Ebert -- nor will I vote for any candidate who solicits my vote via e-mail.
Nick Simicich - njs@scifi.squawk.com
Follow-Ups:
References:
|
|
