Received: from localhost by CS.UTK.EDU with SMTP (cf v2.9s-UTK) id UAA02077; Sat, 26 Aug 1995 20:48:23 -0400 X-Resent-To: drums@CS.UTK.EDU ; Sat, 26 Aug 1995 20:48:22 EDT Errors-to: owner-drums@CS.UTK.EDU Received: from wigate.nic.wisc.edu by CS.UTK.EDU with SMTP (cf v2.9s-UTK) id UAA02070; Sat, 26 Aug 1995 20:48:20 -0400 Received: from VMSmail by wigate.nic.wisc.edu; Sat, 26 Aug 95 19:48 CDT Message-Id: <25082619480682@wigate.nic.wisc.edu> Date: Sat, 26 Aug 95 19:48 CDT From: "Eric Norman (MACC)" Subject: Re: From the Chair: the Reply-To issue To: RAF@CU.NIH.GOV CC: drums@CS.UTK.EDU X-VMS-To: IN%"RAF@CU.NIH.GOV" > > I don't like mailing lists that do that, but I think that they are > > allowed to do so (assuming that they have the original sender's > > permission), modulo concerns that munging the "reply-to" creates a new > > message that really should have a new "message-id". I will argue that they don't create a new message, but that's beside the point of this letter. > How is the mailing list supposed to obtain the sender's permission? > This is something that, when allowed, is usually determined by the list > owner. You could say that by sending to such a list, the sender gives > their permission, but that doesn't clarify anything. In my model of the way things ought to be, there's somewhat of an answer to this question. My model is that an UA normally makes up To: From: Date: Subject: headers, prepends them to the body, and hands that off to the MTA for delivery (I'm well aware than sendmail does some of this UA stuff for you). My model also says that a "normal" reply by a recipient of the message goes back to the address(es) in just the Reply-To: header if one is present, otherwise use the From: header. Now, if a user actually goes to the extra effort to include a Reply-To: header, I think it behooves others to honor it. In other words, if a user includes a Reply-To: header, she is in some sense denying permission. Ergo, if a mailing list is set up for discussion purposes, then the list expander may add a Reply-To: header aiming replies back to the list provided that there is no such header already in the message. [Hmmm, drums@cs.utk.edu is just such a list; its Reply-To: is curious.] I will not go so far as to say that this is required behavior. If a list owner wants to be dogmatic and say "this list is for discussion; replies are going back to the list; period", then so be it; that won't break anything. You know, a lot of this argument really is about default UA behavior. I.e. the argument pretty much amounts to: IF a user wants to send a reply back to a mailing list, do we ask her to (1) Type in "Re:" and subject of message, (2) type in the address of the mailing list, (3) do whatever it takes to include some text from the original message. - or - Just reply "normally" I'm not saying that such arguments shouldn't happen, but the final result should be recommendations that (1) don't break extant behavior, and (2) provide enough information in the message so that authors of UAs can do what they want, or pass on useful options to users. I think the focus should be on the latter case. To that end, it seems to me that the real problem with Reply-To: vis-a-vis mailing lists is that there's no information in the message that identifies it as coming from an exploder. Someone else just said that too; I intend to reply to that message with some of my thoughts. Eric Nnorman (the other Eric)