On Friday 11 August 2006 16:57, Felix Miata wrote: > "Demolished" is mere opinion, not fact.
OK, I'll expand. It is vital for any system that its subsystems conform to its standards. Electricity: BS1362 and BS1363 for the physical connector to the wall (in the UK), CENELEC 230V 50Hz for the signal (in much of Europe). Ethernet: IEEE 802.3. The internet: Internet Protocol. Packets on the internet: Transmission Control Protocol or User Datagram Protocol, amongst others. Email: IETF RFC2822. British roads: drive on the left and follow the highway code. Failure of any actor in one of those systems to conform to the standards is at best troublesome, and at worst catastrophic. I expect my own behaviour and tools to conform to the standards when interacting with any of those systems, and I expect the rest of the system, including the tools used by other end-users and the people running the system itself, to conform to them too. > That cite omits discussion of two key characteristics of > public discussion lists: > 1-Subscribers don't receive messages from authors, they > receive messages from listservs. Wrong. They receive messages _from_ authors _via_ listservers (and relays). The originator of a message is unique; there may be any number of systems which relay / resend / transmit it between its original sending and its final receipt. > 2-The principle of least surprise dictates that public > messages automatically receive public replies in the > absence of special handling by those who wish their > message to go someplace other than from whence it came. Again, the message came _from_ its author, _via_ the list server. The principle of least surprise dictates that when I use the command "reply to author" in my email client, the reply is addressed to the author. If the list has altered the Reply-To address to be the mailing list address (or has added it when it was not originally present), that is unlikely to be the case, since the message headers now erroneously declare that the originator of the message wished replies to be sent to the mailing list. The originator of the message made no such declaration. Reply-To is an "originator field", described in 3.6.2 in the RFC. As such, it must, if present, contain information about the originator (author) of a message, not about one of the systems which processed it subsequently. It's not a "resent field" or a "trace field", both of which may be added by systems performing onward transmission of the message. To quote, "When the 'Reply-To:' field is present, it indicates the mailbox(es) to which the author of the message suggests that replies be sent." Note, _the_author_of_the_message_. So, any mailing list user has the right to place the list address in the Reply-To field, if he or she wants to. The mailing list server does not have that right. If a list server should have the right to make a statement automatically on behalf of the originator of a message, then why shouldn't one of the other systems which transmitted it during its journey from originator to final receiver also have that right? Clearly absurd. > The message each subscriber received did not come directly > from any author, and consequently it should not be > expected to go directly back to any author. I assume "it" should read "replies to it". (Apologies if not.) Put simply, a "reply to author" should go to the author (potentially impossible if Reply-To set or modified by list server). A "reply to list" should go to the list. The behaviour of a generic "reply" command, if present in a given mail client, when the message being replied to contains mailing list headers, should be configurable by a user according to personal preference. It might be addressed to the list, to the author, or to both. Nobody other than the originator (author) of the message has the right to add the Reply-To header, and, furthermore, any system which does so potentially prevents recipients of the message from replying to its originator, should they wish to do so.