[sclug] Key signing

Roland Turner raz.fpyht.bet.hx at raz.cx
Thu Mar 18 16:41:54 UTC 2004


Chris wrote:

> Is it possbile that there is a default behaviour for MUA's and lists.
> Sometimes Reply 
> (Outlook) replys to sender, sometimes to the list. Reply-All often
> replys to the list (if reply goes to individual), or to everyone in
> the last header.

As others point out, this is a contentious issue, but broadly, most mail clients do provide both reply-to-sender and reply-to-all operations. The means of determining the addresses of "sender" and "all" are rather involved[1] and, consequently, are differently interpreted by different implementors and administrators, particularly those who have not taken the time to read the RFC.

When participating in a mailing list, it would be nice to have a reply-to-list option, which is distinct from reply-to-sender and really does not mean the same thing as reply-to-all. As the message headers to even make this possible for MUAs have only come into very widespread use in the last couple of years, very few MUAs offer it. Worse, plenty of MUAs still don't (or only recently started to) offer a reply-to-all operation. Consequently, list administrators have for decades abused the Reply-To header (intended to allow the author of a message to send from one email system but receive replies at another) to coerce replies-to-sender to list messages to be sent back to the list instead of to the original sender. It is my experience that this behaviour is particularly prevelant amongst administrators whose user populations use MUAs which do not have reply-to-all operations, or did not have until recently; PINE springs to mind.

Unfortunately, the practice of abusing the Reply-To header in this way has become so widespread that it has developed a legitimacy of its own. As a result, adhering to the principle of least astonishment has become impossible in that almost as many mailing list participants are sincerely astonished when a reply generated with a reply-to-sender operation does not go to the list as are astonished when such a reply does. This being the case, I see the ideal approach as being to (a) offer the option to each subscriber independently (apparently some mailing list software now does this) and (b) where the subscriber has not specified or can not specify a preference, adhere to the principle of least harm; it is merely annoying to have a reply that you thought you were sending to a list only go to an individual subscriber while it can cause serious harm to have what you thought was a private reply go to an entire list, consequently the default behaviour should still be to leave the Reply-To header alone.

Needless to say, plenty of list admins feel that risking harm to others is an acceptable way of increasing traffic on their super-important mailing list, so the risk remains. When I am subscribed to such lists, I use procmail+formail to undo such cluelessness.

(sigh)

(Did someone say that there are strong opinions on this? :-))

- Raz

1: http://ietf.org/rfc/rfc2822



More information about the Sclug mailing list