Document: FSC-0041 Version: 001 Date: 03/13/90 MSGID / REPLY A proposal for unique message identifiers and reply chain linkage jim nutt 1:114/30@fidonet Information: This FSC suggests a proposed protocol for the FidoNet community, and requests discussion and suggestions for improvements. Distribution of this document is unlimited. From a message originally posted in the NET_DEV conference: The following is a brief description of the proposed standard for generating msgid/reply 'kludge' lines in a message. Msgids have a number of uses besides simply being a unique message identifier. They can be used for dupe killing and reply linkage as well. MSGID A MSGID line consists of the string ^AMSGID: followed by a space, the address of the originating system and a serial number unique to that message on the originating system. i.e: ^MSGID: zone:net/node.point@domain serialno It is not the intent to limit the address field of the msgid to fidonet style addresses, they are used here simply for illustration; messages originating from other types of systems should use an address appropriate to the originating system. The serial number may be anything the developer's little heart desires, AS LONG AS IT IS UNIQUE - NO TWO MESSAGES ON A SYSTEM MAY SHARE A SERIAL NUMBER!! The serial number is formatted as an 8 character hexadecimal integer, i.e. a 32 bit word. Since this yields in the neighborhood of 4 billion unique numbers, I don't think it will be a limit for most systems. A common choice for the serial number is the number of seconds since 1 jan 1970 00:00:00 gmt; this is unique on a single user system and relatively easy to generate. A more elaborate system will doubtless be necessary in the case of a multi-user system. REPLY A REPLY line looks similiar to a MSGID line, and is in fact, very close. REPLY lines are generated using the MSGID line of the message being replied to. The ^AMSGID: string in the MSGID line is replaced with ^AREPLY: and then written to the text of the new message. i.e.: ^AREPLY: msgid of original message Again, this is very simple and as non-compute intense as possible GENERAL For best results, MSGID and REPLY lines should be the first two lines of the message after extended addressing lines (FMPT, TOPT, INTL, DOMAIN), with MSGID appearing above REPLY. This makes it unlikely that a MSGID will be damaged by something that doesn't destroy the binary message header as well. As mentioned above, it doesn't really matter what method is used in generating the serial number, as long as it produces an eight digit hexadecimal number UNIQUE TO THE ORIGINATING SYSTEM. Finally, a MSGID SHOULD BE GENERATED ONLY AT MESSAGE CREATION. IT SHOULD NEVER, EVER BE STRIPPED FROM A MESSAGE NOR SHOULD ONE BE ADDED TO A MESSAGE PASSING THROUGH A SYSTEM (whew, all those caps wore me out!). This is essential for a msgid to be useful, without this restriction, msgids are useless or worse. RATIONALE Finally, what you've all been waiting for (I'm entering this in sixty line mode and the top has scrolled off) ... WHY you should use msgid and reply kludge lines. Good question and one that has several answers, which I will cover one by one... 1) They eliminate the need for origin lines. A msgid contains all the information an origin line is technically there for. It is in a safer place than an origin line and much less likely to be truncated or destroyed while leaving the rest of the message intact. 2) Is related to 1) above, a msgid makes private replies to echomail via netmail a trivial process, you know exactly what system originated the message without worrying about parsing out the origin line (and we all know how much fun THAT is). 3) True reply linkage is possible in echomail because you know exactly which message the reply is to. You just look for a message whose msgid matches the reply in the current message. And you already have a database of msgids for dupe killing (more later). This also lets you track replies ACROSS echomail conferences ... some fascinating possibilities there, eh? 4) Annoying messages can be tracked more accurately. Because msgids are a 'hidden' line and therefore not normally visible or creatable by your average user, they are considerably more difficult to forge than an origin line (forging which is a trivial task). Admittedly this isn't going to stop a twit sysop, but that 8 digit serial number is going to be a problem for him ... if he copies one that already exists, the message will be killed as a dupe, if he makes one up, chances are fairly good that that will be a dupe as well, so it does make forgery a bit more difficult. 5) Accurate dupe killing is possible. By maintaining a database of msgids one can easily check to see if a message is a duplicate of one already entered. If it is stipulated that msgids must be sequential, it becomes, once again, trivial ... simply store the highest serial number that has come from a particular address, any messages that come in with a serial number lower than that can be assumed dupes and killed. 6) Tracking of dupe generators is possible. Since msgids are never stripped and never added, the only msgid on a message should be the one of the originating system. 7) Simplicity. MSGID and REPLY kludgelines are very simple to generate, they require no complex calculations and are easy to parse. And they give you all the benefits above. So why NOT use them? FINALLY Your questions and comments are welcome and will be responded to as time permits. This is only the initial draft, there are a number of proposed extensions to this specification already.