Network Working Group C. Newman Internet Draft: Reply-To-Meaning Proposal Innosoft Document: draft-ietf-drums-replyto-meaning-00.txt November 1997 Reply-To-Meaning Proposal Status of this memo This document is an Internet-Draft. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." To view the entire list of current Internet-Drafts, please check the "1id-abstracts.txt" listing contained in the Internet-Drafts Shadow Directories on ftp.is.co.za (Africa), ftp.nordu.net (Europe), munnari.oz.au (Pacific Rim), ds.internic.net (US East Coast), or ftp.isi.edu (US West Coast). Introduction This is a candidate proposal for one way which the problems with the reply-to header in email could be resolved. Under no circumstances should this be implemented as it is only a candidate for a solution and no decision has yet been made. This proposal distinguishes the different incompatible uses of the Reply-To header with a new Reply-To-Meaning header. This has the advantage of being relatively simple, not invalidating most current practices and allowing mail user agents to present more predictable user interfaces. 1. Conventions Used in this Document The key words "REQUIRED", "MUST", "MUST NOT", "SHOULD", "SHOULD NOT", and "MAY" in this document are to be interpreted as described in "Key words for use in RFCs to Indicate Requirement Levels" [KEYWORDS]. Newman [Page 1] Internet Draft Reply-To-Meaning Proposal November 1997 2. Reply-To Current Practice The Reply-To header is currently used for the following purposes: (1) The author/sender can suggest a complete list of addresses which should receive any reply (e.g., a review committee). (2) The author/sender can recommend an address or addresses to use instead of the "from" address for replies. (3) The author/sender can post to multiple mailing lists and suggest group replies go to only one of them. (4) When the author/sender is subscribed to a mailing list, he can suggest that he doesn't want two copies of group replies to messages he posts to the list. (5) A mailing list can suggest that the list is a discussion list and replies should be sent just to the list by default. Many current MUAs have undesirable results with one or more of these uses. However, if the MUA knew the intent when the reply-to header was added, the undesirable results can be eliminated. Alternative proposals to this one have suggested that the meaning of Reply-To be restricted to (1), (2) or (3) and that additional headers would be added for any of the other uses deemed important. 3. Reply-To-Meaning Header The Reply-To-Meaning header is used to indicate the intent when the Reply-To header was added. It has five values: "any", "private", "group", "non-list" or "list." any The "any" meaning indicates that the author/sender added the reply-to header as a suggested target for any reply (e.g., use 1 above). private The "private" meaning indicates that the author/sender added to Reply-To header as an address for private replies to use instead of the "From" header (e.g., use 2 above). group The "group" meaning indicates that the author/sender added the Newman [Page 2] Internet Draft Reply-To-Meaning Proposal November 1997 Reply-To header as a complete list of addresses where group replies should be sent (e.g., use 3 and/or 4 above). non-list The "non-list" meaning simply indicates that the author/sender added the reply-to header (e.g., any one of 1-4 above). list The "list" meaning indicates that the Reply-To header was added by a mailing list (e.g., use 5 above). The formal syntax using ABNF from message format standard: reply-to-meaning-header = "Reply-To-Meaning:" 1*FWS meaning-keyword meaning-keyword = "any" / "private" / "group" / "non-list" / "list" / extension-token Unknown extension-tokens are treated as equivalent to no Reply-To- Meaning header. Meaning keywords are interpreted in a case-insensitive fashion. 4. Examples Here are a series of examples of various uses: From: Chris Newman To: DRUMS mailing list Subject: I'll collect straw-poll responses Reply-To: Chris Newman Reply-To-Meaning: any In this mailing list posting, the author has asked that all replies go to his straw-poll address. From: Chris Newman To: DRUMS mailing list Reply-To: Chris's DRUMS inbox Reply-To-Meaning: private This author has a special address for private replies to his postings on this list. This might also be used by a someone with a secretary to review incoming responses. From: Chris Newman To: DRUMS mailing list Reply-To: DRUMS mailing list Newman [Page 3] Internet Draft Reply-To-Meaning Proposal November 1997 Reply-To-Meaning: group This author doesn't want private copies of replies sent to the list. From: Chris Newman To: DRUMS mailing list Reply-To: DRUMS mailing list Reply-To-Meaning: non-list This indicates that the author explicitly added the Reply-To header and the mailing list didn't change it. Although meaning (2) can't be distinguished from meanings (1), (3) or (4), this does give assurance that the author will see replies if the Reply-To header is used without the From header. From: Chris Newman To: DRUMS mailing list Reply-To: DRUMS mailing list Reply-To-Meaning: list This indicates that the list is a discussion list and it added the Reply-To header to direct generic replies to the list. There is no assurance that the author is subscribed to the list. In addition, if the author included additional lists or people in a CC header, they probably won't see replies directed only to the Reply-To header. 5. Interpretation in Common Reply Functions Many user agents provide more than one function to construct a default set of target addresses for replies. This section suggests how Reply-To-Meaning can be interpreted by certain common reply functions: Private Reply A private reply is intended for the author(s) or the delegate(s) of the author(s). When there is no Reply-To header, or the Reply-To-Meaning is "group" or "list" then the From header is used as the default set of target addresses for replies. When Reply-To-Meaning is "private", then the Reply- To header is used instead of the From header. If there is no Reply-To-Meaning, or Reply-To-Meaning is "any" or "non-list" then the MUA can offer the user a choice between using the Reply-To or "From" header. Newman [Page 4] Internet Draft Reply-To-Meaning Proposal November 1997 NOTE: if a Reply-To header was constructed with intent (1) or (3)-(5), and was used without warning in a private reply, then the reply could be directed to more recipients than the user intended and cause serious embarassment or other undesirable consequences. Group Reply (also known as Reply All) A group reply is intended for all authors and recipients of the original message. When there is no Reply-To header, it is constructed by using the contents of the original From, To and CC header as the default set of reply targets (with duplicates and possibly the replying user's address removed). When Reply-To-Meaning is "any" or "group" the Reply-To address is used as the complete set of reply targets. When Reply-To- Meaning is "private" or "non-list", the Reply-To, To and CC addresses are used as the default set of targets. Otherwise, Reply-To, To, CC and From are all used as the default set of targets. NOTE: This can result in undesired duplicate copies of a reply when a list address is included which the original author was subscribed to and Reply-To-Meaning is not added. However, this definition assumes duplicate copies are less harmful than risking the omission of the original author. With the opposite assumption, if Reply-To-Meaning was "list" or absent only the Reply-To, To and CC headers would be used. Generic Reply / Gateway A generic reply is simply a desire to reply to the set of recipients most likely to want the reply. It does not specify a preference for a "private" or "group" reply. This is also the set of addresses that gateways are forced to convey when the target mail environment has a single reply-target address list. This is defined as the Reply-To header, if present, and the From header otherwise. User agents which offer this function should draw attention to the Reply-To header when it's present. Reply-To-Meaning has no impact on this function. 6. Mailing List Rules When a mailing list receives a message it MUST NOT alter an existing Reply-To header. If there is a Reply-To header and no Reply-To-Meaning header, the list SHOULD add a "Reply-To-Meaning: non-list" header. Newman [Page 5] Internet Draft Reply-To-Meaning Proposal November 1997 Mailing lists SHOULD NOT add a Reply-To header when one is not present. However, if they do they MUST also add "Reply-To-Meaning: list." In the short term it is particularly undesirable to do this, as it can cause problems for user agents which don't understand the Reply-To-Meaning header. Such user agents may make it difficult to construct private replies. In addition, if a message is posted to several lists, generic replies and some older user agents may fragment the conversation to a single list. 7. Pro/Con Analysis Pros: * This is a fairly simple proposal * This does not invalidate most current practices, which minimizes deployment problems. * This won't cause problems for legacy gateways which will never be upgraded. Cons: * This is an ugly design because it leaves Reply-To overloaded. * Legacy mailing lists which change existing reply-to headers will really mess things up. * This opens up the possibility of permitting lists to add Reply-To headers which could cause short term problems as discussed in section 6. * There is no way to redirect private replies to a non-author and group replies to a different target in the same message. 8. Security Considerations There are a number of cases where a private reply could be misdirected to a large group of people. Although this proposal reduces the chances of this happening, it remains important for MUAs to draw attention to Reply-To headers and reply targets in most situations. 9. Open Issues with this Proposal Newman [Page 6] Internet Draft Reply-To-Meaning Proposal November 1997 The "any" meaning will force current user agents to change their interfaces for private replies when "any" is present. This will probably be controversial. The severity of the pros and cons is probably controversial. 10. References [IMAIL] Crocker, D., "Standard for the Format of Arpa Internet Text Messages", RFC 822, University of Delaware, August 1982. [KEYWORDS] Bradner, "Key words for use in RFCs to Indicate Requirement Levels", RFC 2119, Harvard University, March 1997. 11. Author's Address Chris Newman Innosoft International, Inc. 1050 Lakes Drive West Covina, CA 91790 USA Email: chris.newman@innosoft.com Newman [Page 7]