On 05/Jun/11 18:51, Steve Atkins wrote:
If the point of this is NAT tracking then, AIUI, the meaningful bit of data
is always the source and destination port pair, so maybe it would simplify
things to always record that pair, rather than having two separate records.
Also, IP addresses of either end of the connection are not recorded in
any formally defined manner in trace headers- they're just stuffed into a
comment string in a not entirely consistent or predictable way (try HELOing
with a dotted quad and see what gets recorded where).
Adding formal structure for port numbers where there isn't any for source
and destination addresses seems a bit strange. Either the port numbers
can go in a comment, or the additional structure should record the full
source-ip / source-port / destination-ip / destination-port, maybe?
+1, but this is probably the wrong list, if we expect the I-D authors
to populate their IANA considerations section with the relevant clause
for "Sub-registry: Additional-registered-clauses" of
http://www.iana.org/assignments/mail-parameters