Re: [tlug] getting decent behaviour when hitting 'r' on tlug-mails in mutt

["Stephen J. Turnbull" <stephen@example.com>]

Curt Sampson writes:

 > On 2010-01-10 13:04 +0100 (Sun), Christian Horn wrote:
 > 
 > > Thats the policits-part i dont bring up, but will read up now why 
 > > settings are like this.  
 > 
 > As far as I can tell, it's because the admins don't want replies to
 > the messages they send coming back to both their address and the list
 > address by default (which is perfectly reasonable),

That's awfully generous for someone with a mailbox at cynic.net. :-)
It's not about their personal messages.  The people who initially
invented this abuse were perfectly capable of setting reply-to in
their own messages.  lusers don't even know about Reply-To and
wouldn't think to ask.  The intent was to create a netnews-like
channel that goes out-of-channel for private replies.

The problem is that half-competent implementations of netnews and even
the modern web forums provide for automatically initialized private
replies by email, but this device deliberately destroys the ability of
even a competent MUA the implements all the optional features of the
RFCs to initialize private replies[1], and confuses the user about the
semantics of the commands they use.

In other words, it's fascism: socializing private resources for the
good of the Volk.[2]

 > but don't know how to set Reply-to on messages they send, so they
 > decided just to tweak a global list setting that changes the header
 > for everybody else, too.

No.  It's much worse than that.  This is implemented as a list
setting, but according to the RFCs it's a claim of authorship as well
as an implementation of reply redirection.

Note that it is no longer possible to infer that the author is
strongly discouraging private discussion from Reply-To being set to a
list, nor is it possible to redirect an off-topic or cross-posted
thread to a more appropriate list.  (I personally use Reply-To for
both of those purposes frequently in the context of XEmacs support, to
give you an idea of how that can be very useful in particular
domains.)

 > RFC 2822 e-mail has a way of encoding these author preferences, and
 > that's the Reply-to header (or From/To/Cc when one is not present).

RFC 822 did not fully specify the semantics of Reply-To, in that it
didn't say when Reply-To could be added.  The authors of RFC 2822
debated this, and *deliberately* added the specification that Reply-To
is an "originator header".  This means that someone who adds a
Reply-To header is implicitly claiming to the author or her authorized
representative for mail composition.  Lists which munge Reply-To
therefore are also in principle claiming the right to munge anything
in your mail and to send other mail under your From.

The authors of RFC 5322 (the current Proposed Standard; AFAIK RFC 822
is still STD 11) deliberately did not change this wording.

 > Unfortunately, you have no way of telling what the author really wanted
 > because the list hides that from you. (I believe I've seen it go so far
 > as to change messages where I'd specified a private reply by default
 > to have a public reply by default.)

I don't know about that; ISTR that TLUG has had its munger set to add
TLUG to an existing header field, not replace the header field, as
long as it has been in place.  It is certainly true that other lists
have in the past and probably still do replace an author-specified
reply-to with their own.

The definitive statement on this matter is still "Reply-To Munging
Considered Harmful" by Chip Rosenthal:

http://www.unicom.com/pw/reply-to-harmful.html

There have been more recent additions on both sides, but the defenses
are now moot because they all depend on the assumption that it is
conformant to the mail RFCs, which is it not.  You can get the most
interesting aspects of the history of discussion by Googling "reply-to
munging rosenthal".

A recent proposal to address the need properly is here:

http://turnbull.sk.tsukuba.ac.jp/Blog/Software/ReplyToMungingConsideredEmCarefullyEm

(yes, there is a genuine need for redirection by the mailing list).
Some details need to be updated after discussions with developers of
leading list managers and MUAs, but the basic thrust is unchanged in
recent drafts.


Footnotes: 
[1]  As this thread proves, it is mostly possible to hack around the
list setting, but for those lists that overwrite rather than augment
existing Reply-To fields there is no redress for the author or replier.

[2]  Godwin's law doesn't apply.  There is no charismatic leader on
either side.