Re: Annoying Reply-To

On Fri, 17 Oct 2008, Aidan Van Dyk wrote:

> But now, if the list munged my reply-to, how would you get back to me?

Why'd you have to interrupt a perfectly good, unwinnable idealogical
argument with a technical question? While there is only one reply-to
allowed for a message, you can put multiple addresses in there. It is not
necessarily the case that a list that munges the header must be lossy
(although majordomo isn't a good example here[1]). As most incoming list
messages around only have a from, not a reply-to, you can usefully add
reply-to for regular messages to redirect them to the list (the goal
people who are pro list-based reply to want) and append the list address
to any existing reply-to for the occasional odd message that specifies it
directly, like yours I'm replying to.

As for an actual implementation of good behavior here, see the end of
http://www.gnu.org/software/mailman/mailman-admin/node11.html for one
example of list software that supports adding a reply-to without stripping
any already there off in the process.

From a RFC 5322 standards-based perspective, I see the crux of the
argument like this: the reply-to is supposed to be set to the
"address(es) to which the author of the message suggests that replies be
sent". The RFC says the author is "the mailbox(es) of the person(s) or
system(s) responsible for the writing of the message". I don't think it's
completely unreasonable to say the system running the mailing list
originating the actual message into my account could be considered a
co-author of it by that definition. It's a system with a mailbox that's
responsible for me receiving the message, and the fact that it does touch
some headers says it's writing part of the message (if you consider the
header part of the message, which I do).

I'm OK that such interpretation is not considered correct, but it does bug
me a bit that most of the arguments I see against it are either strawman
or appeal to authority based rather than focusing on the practical.
You'd be better focusing on real-world issues like "oh, if reply-to were
set to the list, every idiot subscribed with an auto-reply that doesn't
respect the bulk precedence would hit the whole list, thereby introducing
the potential for an endless mail-loop"; now that is much easier to
swallow as a problem for a large list than RFC trivia.

[1] majordomo doesn't handle this very well out of the box as far as I
know. I believe its only behavior is still to only replace the reply-to
with the list address. You can insert something in between where messages
are approved as list-worthy (attachments aren't too big, etc.) and when
"resend" is called to implement the same feature: add or extend the
reply-to with the list address while not losing any explicit reply-to in
the original. But I think you still have to hack it in there yourself, as
I did once long ago.

--
* Greg Smith gsmith(at)gregsmith(dot)com http://www.gregsmith.com Baltimore, MD