[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ISSUE #1414: Call for consensus




--On 21. mai 2007 16:33 +0000 Charles Lindsey <chl@xxxxxxxxxxxxxxxx> wrote:



In <87irapo8oz.fsf@xxxxxxxxxxxxxxxxxxxxx> Russ Allbery <rra@xxxxxxxxxxxx>
writes:



--- usepro.xml	2007-05-18 17:08:35.000000000 -0700
+++ usepro-1414.xml	2007-05-18 17:26:33.000000000 -0700
@@ -373,16 +373,22 @@
+              <t>The agent MAY then prepend to the Path header field
+              content "!" or "!!" followed by an additional
+              &lt;path-identity> for itself other than its primary one.
+              Using "!!", and thereby adding a &lt;diag-match> since the
+              &lt;path-identity> clearly is verified, is RECOMMENDED.
This +              step may be repeated any number of times.  This is
permitted +              for agents that have multiple
&lt;path-identity>s (such as +              during a transition from one
to another).  Each of these +              &lt;path-identity>s MUST meet
the requirements set out in +              <xref target="path" />.</t>
+
+              <t>Finally, the agent MUST prepend its primary
+              &lt;path-identity> to the Path header field content.  The
+              primary &lt;path-identity> is the &lt;path-identity> it
+              normally advertises to its peers for their use in
generating +              &lt;path-diagnostic>s as described above.</t>


Yes, I think that is all agreed.

There remains the related ISSUE of whether agents may OMIT some entries
from the Path (e.g. when they merely record some inner communications
within the one site). But the proposed alteration to allow that affects a
different (earlier) piece of wording.
If I were being pedantic, I'd claim that a system of entities that adds one and only one entry to the path must be regarded as one agent, no matter how many computers or intermediate communication hops it consists of.
In the same way that an (IP) router will decrement the TTL of IP packets passing through it; a box or set of boxes with a summary function that decrements the TTL by 1 will have to be regarded as one (IP) router.
Actually, I don't see a reason to depart from pedantry at this point. Seen in that light, the "remaining issue" can't be stated in a meaningful way, so it must be regarded as resolved. 

                Harald