Is cpim-srv document necessary at all? (Re: NAPTR and CPIM)
Pekka Pessi <[email protected]> 19 Nov 2002 08:09:55 +0200
| Newsgroups | gmane.ietf.impp |
|---|---|
| Message-ID | <[email protected]> |
Hello all,
Based on today's discussion in IMPP I now have came to the
conclusion that the SRV lookup does no good (slow thinker, that
I am).
If a client can send the IM using the protocol that they have
picked, they send the IM using it. Otherwise, they use a
preconfigured gateway. In any case, they don't need a
CPIM-specific DNS lookup but they will use the protocol-specific
(SIMPLE, XMPP, Jabber, AIM, WV, what ever) method to resolve IM
URIs.
This has problems, however:
1) When we know that the destination domain does not support our
protocol of choice? Is it enough that there is no NAPTR/SRV
records (or whatever) for our protocol, or should we try to
resolve an AAAA record and send IM to the address there? What if
the destination domain supports only legacy IP, should we
resolve also A record and try send IM there, too? Who should try
to do send IM to AAAA/A? Client? Gateway? Both? Do they try each
IM protocol that they know?
This problem applies to the current SRV-solution, too. A client
or gateway simply cannot lookup SRV record for *all* possible
CPIM protocols
(I should mention here that if NAPTR was used, you could get a
list of all IM protocols supported by a domain. If the DNS
server was up-to-date, the response message to the initial NAPTR
lookup would also contain SRV records *and* corresponding AAAA/A
records of the preferred IM service.)
2) The destination domain supports our protocol of choice, but
how we know if it supports IM URLs using that protocol? For
example, the destination domain uses SIP for telephony but
Jabber for IM.
I really don't know if this is a usual or even reasonable
scenario, but in any case, I think this is a problem specific to
the protocol, not a problem for CPIM.
So, I propose that CPIM just settles on a solution for problem
#1, and leaves the problem #2 to be solved by the protocols in
question.
Of course, this would be bad news for CPIM editor Jon Peterson
(no more srv document? what I'm going to do with all this free
time) but good news for SIMPLE chair Jon Peterson (kewl, yet
another document to charter).
Pekka
[reminder: [email protected] for non-technical discussions, please]