From: Greg Berigan (gberigan@cse.unl.edu)
Date: Thu Sep 04 1997 - 20:49:53 CDT
Brad Templeton <brad@clari.net> wrote:
>Greg Berigan <gberigan@cse.unl.edu> wrote:
>> Field? It's just a string. It is data, not a token standing for some
>> other string.
> Right, I'm saying we need to make it an official token that is understood
> by software that might issue replies, so that software knows to not
> attempt to reply, and instead issue an error.
Then we apparently agree. I noted the historical treatment of the nobody
user as an indication that this won't seriously break legacy software. Old
software would try to mail to a local null user, but new software would
recognize "nobody" and, as you say, issue an error rather than even
attempting to mail. It just has limited backward compatibility. People
really could use it right now without serious problems.
>> talking about "Reply-To: nobody" to prevent replies. It makes just as much
>> sense as "Followup-To: poster", IMO.
> Well, I view "followup-to: poster" as an indication that the poster would
> prefer to receive private replies, and not a declaration that the user
> can't do a followup if the user desires it. However, I do not see a
> need to allow an ordinary poster to prevent replies, I only see a need
> for that on postings by moderators.
It serves to discourage private responses to keep discussion public, as
well as to discourage carbon copies of messages. The knowledgeable user
can always override this manually just as they can bypass Followup-To:
poster. It has always been up to the implementors to define the user's
options in this case; nobody won't be any different.
> I would consider the preventing of replies by an ordinary poster to be
> extremely rude, and would almost surely killfile such messages. If you
> don't have the time to read my reply, I don't have the time to read your
> message.
And what about people who post their question and then want e-mailed
replies only (typicaly not setting Followup-To: poster)? I'd argue that
that is also extremely rude, possibly moreso. It seems the only polite
behavior is to allow the person responding decide, but I am sure that there
are some good reasons the poster would have in setting them, just as
alt.dev.null serves a purpose.
> That is of course just my opinion, so I would not want to
> force people to follow such a rule. As such an official address that
> means "this is not an address" should be defined, and software should
> be able to recognize it. Because some people will wish to stylize such
> addresses, I advise a pattern rather than a single address.
Such wishes to stylize their own null addresses is not something we need to
grant. We might as well let everyone create personal vanity newsgroups
like and use them instead of poster in Followup-To.
There isn't much we can do to prevent either method. I just think we
should have a supported method to do this that everyone can use in the
standard so we don't have bogus addresses which will follow no pattern
appearing in Reply-To. (Some poeple are already using valid froms with
munged reply-tos which make no sense.)
> Having
> a domain is necessary to make the address conform and to not have it
> conflict with local addresses. We surely are not going to allow
> "Reply-to: postmaster" as a valid line, which would direct replies to the
> local postmaster at each recipient site...
Why not? How about a message which says, "If your ISP doesn't have
procmail or you don't know if it has procmail, you can reply to this
message and it will be sent to your local postmaster who will be able to
answer your questions." Note that this is _not_ equivalent to
postmaster@[127.0.0.1].
Again, the fact that old software will send mail to the local nobody user
has no bearing on what new software will do, which will throw up an alert.
And I'm still against invalid address across the board. If someone needs
to be anonymous to the world, they should find a remailer or other person
to front a valid address for them. Any MAY or SHOULD NOT for munging will
be validation and official acceptance of this behavior, even if there is a
MUST on the flagging. I'm dismayed it is even being considered.