Received: from localhost (daemon@localhost) by CS.UTK.EDU with SMTP (cf v2.9s-UTK) id SAA17226; Thu, 21 Mar 1996 18:53:47 -0500 Received: by CS.UTK.EDU (bulk_mailer v1.4); Thu, 21 Mar 1996 18:53:12 -0500 Received: from koobera.math.uic.edu (qmailr@KOOBERA.MATH.UIC.EDU [128.248.178.247]) by CS.UTK.EDU with SMTP (cf v2.9s-UTK) id SAA17163; Thu, 21 Mar 1996 18:53:09 -0500 Received: (qmail-queue invoked by uid 666); 21 Mar 1996 23:55:13 GMT Date: 21 Mar 1996 23:55:13 GMT Message-ID: <19960321235513.4808.qmail@koobera.math.uic.edu> From: djb@koobera.math.uic.edu (D. J. Bernstein) To: drums@cs.utk.edu Subject: Re: Message-IDs (not Received lines) (Was: Issues summary web page) > And even in the case of bounce messages, they are > especially useful for dealing with duplicate messages: I disagree. The bounce message has no useful identity except in its role as information about the original message. If a bounce is duplicated, you end up learning twice that the original message was not delivered. > If you are sent > multiple copies of a message by some broken piece of software, especially > bounce messages, you can just automagically delete ones with duplicate > message-ids. This is insecure and must not be implemented. To automatically delete a duplicate without compromising security, you need to compute a one-way hash of an appropriately stripped message; the Message-ID doesn't help. > There is no reason not to generate them, and they are quite useful. I am worried about the motivation for, and likely consequences of, requiring them. ---Dan