From: Ofer Inbar <cos@leftbank.com>
Message-Id: <199612100119.UAA24666@zax.leftbank.com>
Subject: Source Routing confusion
To: drums@cs.utk.edu
Date: Mon, 9 Dec 1996 20:19:18 -0500 (EST)
Organization: The Left Bank Operation - http://www.leftbank.com/
Reply-To: cos@leftbank.com
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

I guess several of us are confused about the treatment of source
routing in the current 821bis draft, and John is planning to rewrite
it.  So rather than a long discussion, here's a brief list of what I
am finding confusing...

I.
  In 3.7, third paragraph says "If source routes are used, RFC 821
should be consulted for the mechanisms for constructing and updating
the forward- and reverse-paths."  The fourth paragraph then describes
how to updated the forward- and reverse- paths.  Was this intentional?

<aside>
  If there's a backreference to RFC821, and then text in the current
document describing how to do the same thing, which takes precendence
in the case of a disagreement between the backreference and the
current document?  Not to say there's a disagreement here - I haven't
looked at how 821 deals with source routes, but in general...
</aside>

II.
  In 3.7 there's an example of an undeliverable mail notification,
illustrating the use of a null reverse-path.  Fortunately(?), this
example also shows a source route in the forward-path, which led me to
think about how the rule described in the fourth paragraph of 3.7
would apply to this example.  Following that rule, shouldn't
 MAIL FROM:<>
 RCPT TO:<@HOSTX.ARPA:JOE@HOSTW.ARPA>
become:
 MAIL FROM:<@HOSTX.ARPA:>
 RCPT TO:<JOE@HOSTW.ARPA>
at the next hope?

Clearly that isn't what we want, but it follows from the rule.
Or maybe not so clearly.  Just how is a null reverse-path special?

III.
  Source routes are again mentioned in the section on RCPT, 4.1.1.3.
"... relay host SHOULD strip or ignore source routes and names MUST
NOT be copied be copied into the reverse-path."  There is an example
in this section:

  For example, mail received at relay host A with arguments
    FROM:<USERX@HOSTY.ARPA>
    TO:<@HOSTA.ARPA,@HOSTB.ARPA:USERC@HOSTD.ARPA>
  will be relayed to host B with arguments
    FROM:<USERX@HOSTY.ARPA>
    TO:<USERC@HOSTD.ARPA>

Will be relayed to host B?  That's what the source route said to do,
and that's what the text says SHOULD be stripped or ignored.  Is the
example meant to defy the SHOULD, or was this a typo?  And if it was a
typo (or a mistaken copy from RFC821), what about the fact that
"fixing" it by saying "relayed to host D" would contradict section 3.7?

I suppose this could just be a case of "examples not yet revised",
but either way, there's still a conflict between 4.1.1.3 and 3.7.

  --  Cos (Ofer Inbar)         --  cos@leftbank.com cos@cs.brandeis.edu
  --  The Left Bank Operation  --  lbo@leftbank.com http://www.leftbank.com/
  --  WBRS (100.1 FM)          -- WBRS@brandeis.edu http://www.wbrs.org/
       "Aren't you ever *not* on-line?"   -- Caitlin Xantha Hazen