Russ Allbery wrote:
Bruce Lilly <blilly(_at_)erols(_dot_)com> writes:
It is harmful because it results in an increase in spam; e.g.:
"Hi. This is the qmail-send program at cr.yp.to.
I'm afraid I wasn't able to deliver your message to the following addresses.
This is a permanent error; I've given up. Sorry it didn't work out.
<djb-internal/copies(_at_)cr(_dot_)yp(_dot_)to>:
See http://cr.yp.to/docs/copies.html.
That has nothing to do with MFT and everything to do with Dan Bernstein
not being interested in you sending him copies of mailing list mail.
You are mistaken on both points: the URL above is a rant page specifically
about the non-standard "mail-followup-to" field. And I have not sent
Bernstein copies of mailing list mail, as you could easily have determined
from the mailing list copy; the message was sent *to* Bernstein with a *copy*
to the list. And that is exactly what would be expected from a response
to a message (his) which contained no Reply-To field indicating that
responses should go anywhere but to him (with reply-all copying the
list). As specified for the semantics of Reply-To and responses in
RFC 724 [sect II.C.. paragraph 3)], RFC 733 [IV.A.2.c], RFC 822 [4.4.3],
RFC 2822 [3.6.2], and RFC 3834 [sections 3.1.2 and 4]. None of which
mention "Mail-Followup-To".
That is, incidentally, something that you do on all the mailing lists that
you're on, and it's incredibly obnoxious. I haven't gotten to the point
of bouncing your mail back to you yet, but I wish you'd quit it too.
Like many mailing list participants, I use a "reply-all" function to
send list responses unless there is a good reason to do otherwise, and
generally trimming extraneous recipients (typically recipients who had
been copied on earlier list messages, but for whom the response being
composed isn't necessarily relevant). There are several typical cases
of mailing list messages that might be responded to:
1. Reply-To set to the list address, no To or Cc. This is typically
how an author would indicate that responses should be sent to the
list only. In some cases, multiple addresses might be specified
in the Reply-To field.
2. Reply-To not set, a single author's single mailbox in a From field,
To or Cc set to a list mailbox That indicates that the author
wishes to receive responses (reply or reply-all) and that reply-all
responses should copy the list address.
3. Reply-To set to the list address and one or more additional mailboxes
specified in To and/or Cc fields. A "reply" response would go to the
list only, whereas reply-all would go to the list and the additional
mailboxes. This typically arises due to a response to type 2 above,
i.e. where the mailboxes in the type 3 message had been indicated in
the earlier type 2 messages as requests for copies.
4. Reply-To not set, a single author's single mailbox in a From field,
list mailbox in either To or Cc, and additional mailboxes specified
in To and/or Cc. This may occur as a response to a type 2 message
where the respondent wishes to have responses to his list message
sent to him, with reply-all responses sent to him, to the list, and
to the additional mailboxes specified.
5+. In the case of multiple authors, a Sender field is necessary regardless
of To/Cc/Reply-To fields. Because of different interpretations of
the Sender field, it would be advisable for the authors to include a
Reply-To field. This is a very unusual situation; I have only seen
it a few times.
If you are receiving direct responses as well as mailing list copies,
and you didn't intend that to be the result, you are probably sending
type 2 or type 4 messages when you probably should have been sending
type 1 or type 3 (i.e. you failed to include a Reply-To field indicating
that responses should be sent somewhere other than the mailbox in your From
field).
Responses to mailing lists aren't magical or mysterious; they are handled
in accordance with the specified semantics referenced above. There is
really no viable alternative (at present, hence this thread). You seem to
want to be magically excluded from responses without simply specifying
(via Reply-To) where you want responses to go instead. In theory,
It might have been possible in pre-spam days for a response composer to
a. find the MX host(s) for the mailing list mailbox
b. telnet to the MX hosts(s), issuing an EXPN SMTP command for the list
mailbox
c. save the results
d. find the MX hosts for each specified message recipient
e. telnet to each MX host, again issuing EXPN SMTP commands, and again
saving the results
f. recursively process the saved results, again issuing SMTP EXPN commands
to appropriate MX hosts until no further expansions result
g. compare results of the expansions and eliminate recipients who would
be included in the lists expansion, any subsequent additional expansions,
aliasing, etc.
Nowadays, EXPN mostly doesn't work, so that can't be done. Hopefully
you can see that it it impractical, even if it were possible. At minimum,
the mailing list subscriber list might change between the time it is first
expanded and the time that the recipient list is edited. Ditto for any
intermediate expansions and aliasing. There simply is no way for a list
recipient (who is not also a list maintainer) to determine who is and
who is not a list recipient. Specifically, I don't know which mailing lists
you receive at any given moment. And to address your assertion, you don't
know "all the mailing lists" that I am subscribed to at any moment. The
difference is that I don't expect you to know or to jump through hoops
in a futile attempt to find out.
Moreover, as has been recently stated in this discussion, at least one
mailing list recipient who sends type 2 messages *does* expect responses
sent to him with copies to the list. Since that is precisely the expected
sort of response to a type 2 message, the only way *you* can expect
anybody else to know that *your* wishes are different is to so indicate,
preferably by using the standard, well-documented Reply-To field, which
has been available for that purpose for more than 27 years.
What is really obnoxious is the apparent expectation that every list
recipient will go through steps a-g above to suit senders who can't be
bothered to set the Reply-To field on their list messages if they don't
wish to receive responses. That isn't going to happen and shouldn't be
expected -- clearly this is another case where some user education is
needed.
And to return to the topic at hand, the sort of magic that you seem to
expect won't happen (i.e. the From mailbox will still receive responses
in the absence of a Reply-To field) even *with* Keith's proposed noReply
fields.