From: Paul Kyzivat <[EMAIL PROTECTED]>

   [EMAIL PROTECTED] wrote:

   > We still need to investigate what History-Info looks like if some of
   > the participating proxies do not support it.

What happens with History-Info if a proxy does not support it, a proxy
that is between the most-upstream proxy that adds History-Info and the
agent that is interpreting it?

There seem to be two messy aspects:

1) There must be an unambiguous way for the SIP agent to determine the
History-Info element that is the leaf that applies to it.  In
principle, there should be only one leaf whose URI matches the current
request-URI.  But in SIP, "address matching never works".  And if
there can be intermediate proxies rewriting the request-URI but not
updating History-Info, address matching is hopeless.

One solution is to rule that the element with the "rightmost" index is
always the applicable element.  This can probably be made to work, but
it requires that even when there is parallel forking, no fork can see
another fork until that fork has returned a final response.

Another solution is to specifically mark the applicable element with a
field-parameter.

2) The intermediate proxy can fork the request.  This results in two
or more downstream agents adding History-Info elements with the same
indexes.  Fortunately, only one of these gets back to the upstream
proxy, so from that proxy's point of view, assembling the History-Info
is unambiguous.

However, if a HERFP mechanism allows the upstream proxy to see more
than the History-Info for more than one fork under these
circumstances, it would have to integrate the History-Info's and may
have to renumber the elements.  Messy but possible.

   Right! More importantly, the middle boxes that are intentionally trying 
   to obscure it. It seems that it is common to want to hide all the 
   details. I find it very unlikely that HI will ever be populated in a 
   reliable enough way to be of use in making important decisions.

That's only a problem if History-Info from one side of an SBC is
needed to make a decision on the other side of the SBC.  Or, if
History-Info must pass through a topology-hidden network, we would
have to ensure that the SBCs cooperate to transport the information
through the network.

Dale
_______________________________________________
Sip mailing list  https://www.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [EMAIL PROTECTED] for questions on current sip
Use [EMAIL PROTECTED] for new developments on the application of sip

Reply via email to