msgtrk-mtqp-00 comments

Gregory Neil Shapiro <[email protected]> Fri, 28 Jul 2000 13:06:49 -0700 (PDT)
Newsgroups gmane.ietf.msgtrk
Message-ID <[email protected]>
The only document author is Tony but the page breaks list "Hansen, Allman".

2. Basic Operation
   - s/MTQP SERVER SENDS A GREETING/MTQP server sends a greeting/
   - s/10 minutes' duration/10 minutes in duration/  (??)
   - Beyond an inactivity timeout, do we want to mention something like:
     "An MTQP server MAY limit the number of commands or total connection
      time to prevent denial of service attacks."

4. Track Command
   - The response is described as a "[MIME] mail message".  I would prefer
     "[MIME] formatted part".  To me, a "mail message" includes a
     Message-ID:, Date:, Received:, etc. headers.  A MIME part does not.
   - We should have an example for the chaining response case:

     Example #XXX
     C: TRACK <tracking-id> 1234567890ABCDEF
     S: +OK+ Tracking information follows
     S: Content-Type: multipart/report; report-type=tracking-status;
     S: 	boundary="1234567890ABCDEF"
     S: 
     S: --1234567890ABCDEF
     S: Content-Type: message/tracking-status
     S:
     S: ... server1 details go here ...
     S: 
     S: --1234567890ABCDEF
     S: Content-Type: message/tracking-status
     S:
     S: ... server2 details go here ...
     S: 
     S: --1234567890ABCDEF--
     S: .

9. IANA Considerations
   - Beyond "registered extensions", we also need a registered service name
     and port number.

10. Security Considerations
   - Since we are separating the protocol from SMTP, I think we should rely
     on only the security considerations of the other drafts.
     Specifically, this protocol has unique security considerations (see my
     next item).
   - Should mention use of an encrypted SMTP session, otherwise the
     tracking request is vulnerable to replay attacks such that outsiders
     who snoop a request can later get more information by sending the same
     information again (given that we don't have a shared secret to do APOP
     style authentication).

11. Protocol Syntax
    - There should be a blink line between the first text line and the
      beginning of the ABNF
    - There should be a newline before the second ';' comment character in
      the multi-char-success ABNF