Two-replies problem and Reply-To: munging (was: Today's meeting! ... and ...)

[Rick Moen <rick@linuxmafia.com>]

Quoting Michael Paoli (Michael.Paoli@cal.berkeley.edu):

> Yes, there's meeting today!:
> https://berkeleylug.com/meetings/
> And, if it's like the last couple or so,
> I believe there will also be in-person - for those that
> can make it and are and feel safe to do so, etc. (see the
> earlier posts).
> Alas, I've got a scheduling conflict, so will miss at least
> part of the meeting ... maybe even all of it.  :-/
> But I may catch part of the in-person - presuming that's happening
> today.
>
> Anyway, hope all have a fine meeting in all cases!

Yes, I'm pretty sure we solved all of the world's problems -- but,
tragically, forgot to take notes, so we'll just be obliged to meet
again.  ;->

During the meeting, Thomas asked how to send on-list replies on
sf-lug@linuxmafia.com (and impliedly, though he didn't phrase it that
way) to other old-school mailing lists where the MLM (mailing list
manager software that runs thing) doesn't "munge" (force) the Reply-To: 
header.  Thomas uses the GMail WebUI, which reportedly _does_ present to
the user separate reply-sender (often called just "reply") and reply-all
user controls.

Reportedly, GMail's WebUI doesn't do any intelligent handling of mailing
list replies like some mailers.  It doesn't have a reply-list mode, or
automatically trim non-mailing-list recipients when the user does
reply-all.

The simple answer to Thomas's question was:

o  Always use reply-sender to send direct, private mail just to the poster, and
   not to the mailing list.
o  Always use reply-all to send a public reply to the mailing list forum.

That is _supposed_ to do the intended thing everywhere.  That is the
intended purpose.

SHORT VERSION:  So, do that.  No, the results aren't perfect.
Controversy over related matters was ended by IETF (Internet Engineering
Task Force) in 2001, and that's what the Internet standards say.



LONG VERSION:

On some mailing lists such as berkeleylug@googlegroups.com, the MLM
software has been configured to add a "forcing" ("munging") header to
every message the subscriber sends, e.g., BerkeleyLUG's mailing list adds 
Reply-To: berkeleylug@googlegroups.com

The logic of making the MLM do that is as follows:  (1) Subscribers
hardly ever want to send offlist mail, but almost always want to
continue the public thread.  (2) Subscribers shouldn't need to learn two
separate reply commands.  (3) So, let's hijack the Reply-To header and
force it (via the MLM) onto all postings, so that users can just hit 'r'
and the result will be a public reply.


There are two major problems with that approach.  Problem 1: That's not
what the Reply-To: header is for.  It's for the sender's use to state a
preferred reply address where he/she would prefer to be reached.  For
example, imagine I'm about to go on a three-month vacation at a cabin in
the woods, in Inverness, and that while there I'll be reachable at
rickmoen@inverness.com .  I might thus, for the week or two before
vacation day, sent out mail with:

From: Rick Moen <rick@linuxmafia.com>
Reply-To: Rick Moen <rickmoen@inverness.com>

If a mailing list overrides this breadcrumb I included in my mail and
hijacks Reply-To: for some other purpose, then it is interfering in the
header's legitimate function.

Problem 2:  Worse, it also interferes in what a mailer's reply-sender
command is supposed to do, by forcing "munging" the communication to
public discussion when the sender tried to send a private mail.  If the
contents are seriously private, the result could be embarrassent, a
bloody nose, a divorce, a termination of employment.

The fact of the matter is, there's a good reason why there are two
different reply commands, and forcing the reply-sender action to do the
work of reply-all, _just because_ some users want to learn only one
reply command an not two, is pigheaded and does harm.

Accordingly, no mailing list I administer does "munging".


During today's meeting, I inquired if anyone present administers a
Google Group, as I was pretty confident Google Group permits either
standards-compliant operation or "munging".  People seemed to doubt this.
I just now looked it up, though, and I'm _correct_.  To set
standards-compliant operation (no "munging"; leave subscribers' headers
alone), if one is Manager of a Google Group:"

1. Open the Group.
2. Click Settings, choose "Group Settings".
3. In section Settings, choose "Email Options".
4. In field "Post Replies" choose "to the author of the message only"
   on dropdown menu.
5. Save.

In other words, BerkeleyLUG's mailing list at Google Groups doesn't
_need_ to do "munging".  Someone in charge has made it do that.
(One of the many reasons I love the mutt mailer is that it can be easily
configured to ignore "munging", so mutt users can avoid being hurt by
it.)



Thomas, who's polite, observant, and concise, said (paraphrased) "But
when I do reply-all in GMail to a mailing list like
sf-lug@linuxmafia.com, doesn't that result in an undesired direct
private-mail reply to the previous poster, in addition to the on-list
public response?"  Yes, it does, because GMail's WebUI apparently
doesn't do intelligent processing of replies to mailing lists 
(as do, for example, mutt and Emacs GNUS).

So, what's up with that?

Let's say I posted to SF-LUG:

From: Rick Moen <rick@linuxmafia.com>
To: sf-lug@linuxmafia.com

Let's say Thomas is also a subscriber, and does reply-all.  By definition,
"all" means both of the two cited addresses, and there is nothing
especially special about a mailing list address in the SMTP standards.

From: tom r lopes <tomrlopes@gmail.com>
To: Rick Moen <rick@linuxmafia.com>, sf-lug@linuxmafia.com


So, that's why; because all means all.  It's possible to have a mailer 
that improves on that, the way mutt and GNUS do, but GMail's WebUI
doesn't.


Anyhoo, back in the last century, there was an infamous, noxious, and
endless Internet flamewar about Reply-To: "munging", and I carefully
avoided participating but just FAQed the matter.  Then IETF settled the
argument in 2001 with RFC-2822.  The "munging" position lost.  The end.


Except, sadly, here we are 20 years later, and some people still think
they want "munging".  In 2006, Neale Pickett had a nice essay, recapping
the history briefly and how it is now a dead issue.  Marc Merlin hosts a
mirror copy:
http://marc.merlins.org/netrants/reply-to-still-harmful.html Please see.


Thomas might find interesting Pickett's brief bit about the "two
copies" problem:

  Getting two copies of the same email

  Some people complain that they'll get two copies of the same email.
  Since they're on the list, their first copy is the one sent to them by
  the list.  When the responder hit "reply all", it also put their email
  address in the recipient list, so they get a second copy directly.

  Fortunately, there's already a technical solution to this.  Since all
  mail clients put a unique Message-ID header on their email, a mail
  reader has only to compare the Message-ID of a message to
  previously-received messages. If it's the same, then the second message
  is a duplicate and can be safely ignored.

  If your mail reader doesn't do this, that's too bad, but it's not an
  excuse to violate Internet standards and surprise people with
  inconsistent behavior, just to prevent you from having to delete a few
  emails.  Anyone who gets any spam at all knows how to delete email.



-- 
You received this message because you are subscribed to the Google Groups "BerkeleyLUG" group.
To unsubscribe from this group and stop receiving emails from it, send an email to berkeleylug+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/berkeleylug/20210614062501.GP12541%40linuxmafia.com.