SIPPING Session 2 Notes

Reported by Spencer Dawkins


WEDNESDAY, March 3, 2004, 1300-1500, Crystal 3
1300 Agenda Bash
Chairs
Design Teams
1305 Application Interaction
Jonathan Rosenberg
draft-ietf-sipping-app-interaction-framework-01.txt

- Correlation SUBSCRIBE with dialog - was using GRID, but doesn't work
with forking - must have unique GRIDs
- remaining open issue - how to match REFER to associated component?
- can't send REFER on same dialog - many known problems encountered
with SUBSCRIBE, etc.
- proposal is "no correlator" - this means web app won't automatically
close when call ends
uses "what we do today" for apps
Rohan - have proposed new parameters on REFER-TO
can't authorize without explicit correlation
Cullen - we can't count on closing browser
- we actually need an explicit correlator - no objection

- Reg Event - unable to add path information with current version in
RFC editor queue
adding a <uri> to <contact>
- Do we still need "params", now that we have <uri>?
alternative is to define specific values instead of use
parameters, but what if we have an extension that's not defined?
alternative is to use XML extension, so reg event notifications
won't contain data unless the server understands it
SIMPLE has the same problem with prescaps
get this right - it's an AUTH48 change and the RFC will appear
immediately
Rohan - <unknown name="whatever">
what if a parameter can have more than one type?
two types, token and string, but types are different and
comparison rules are different - can't do this without typing
need to come up with a proposal and post it to the list

KPML
Eric Berger
draft-ietf-sipping-kpml-02.txt
- Does KPML subscribe to what the phone would send or to what was
sent?
Maybe both
- Notify from UAC to UAS, proxy, or server
- small changes from 01 draft, mostly additions
- Should I use GRUU or not? would interoperate in all scenarios, but
requires UAS registration and GRUU support
no objection, so will use GRUU

1315 SIP Conferencing
Alan Johnston
draft-ietf-sipping-cc-conferencing-03.txt
draft-ietf-sipping-conference-package-03.txt
draft-ietf-sipping-conferencing-requirements-00.txt
draft-ietf-sipping-conferencing-framework-01.txt
- no remaining document holdups
- some changes to conference package in last cycle
- added requirement that cc-conferencing conference URI must be a GRUU
nicknames are more than a conferencing thing, or even a messaging
thing - we need to solve this in SIP, not in conferencing
we've made it clear that display names are no more than nicknames,
and certainly not an identity
we've had requests to change display names before
but how do vanity URIs work with GRUUs?
- documents almost ready for WGLC, with example document still to be
written by XCON
high level requirements and framework still need to be updated

1325 Transcoding
Gonzalo Camarillo
draft-camarillo-sipping-transc-b2bua-01.txt [PDF]
- Should we be doing Route headers? Seems like a stretch, we usually
use Route to talk to a proxy
- This discussion ties to Session Border Controllers
- we don't have consensus whether a conference is a B2BUA or a proxy?
- can we stick a transcoder in a route header?
a B2BUA actually terminates a call - is this what we want?

1345 Location Conveyance Requirements
James Polk
draft-ietf-sipping-location-requirements-00.txt
- no known open issues for U2U location conveyance
- emergency calls do have open issues
always require locations, whether reliable or not?
flag some locations as "possibly wrong"?
accommodate VPNs?
include indicator of where the location information came from
(GPS, triangulation, ...)?
is this quality, preciseness, or something else?
is this who knows the information or where it came from?
want to be able to figure out, post-mortem, how to figure out why
a specific location was provided to an ERC
proxies should be able to insert location information for an
emergency call?
should we require a self-signed cert for S/MIME?
- recommends document rev, including solutions, by San Diego

1335 SIP Emergency
Henning Schulzrinne
draft-ietf-sipping-sos-00.txt
draft-schulzrinne-sipping-emergency-arch-00.txt
- sos draft has been changing
- support emergency calls in SIP for ERCs that are SIP-enabled
- need security and privacy
location insertion by UA
- trying to roughly align with 3GPP
- thinking of DNS sos.arpa mechanism (but this is not required)
call routing by UA?
- Location update via INFO, or something else? - not call-related
- refuse caller hangup (in some jurisdictions) - refuse "403 BYE"
2833 tone?

1355 REFER Extensions
Orit Levin, Rohan Mahy, and Gonzalo Camarillo
draft-olson-sipping-refer-extensions-01.txt
draft-mahy-sip-remote-cc-01.txt
draft-camarillo-sipping-multiple-refer-00.txt
- How to use REFER to INVITE multiple people to a conference (also
BYE)
this can be done using other means - do we want to do it?
proposal is: not at this time
- Replace Refer-to URI syntax with a MIME body?
could place cid in the body, in XML format
- Could surpress implicit subscription
"norefersub" - no extension required
- use arbitrary event packages with REFER?
need to polish syntax at a minimum
allow this for any event package?
- REFER SIP responses with codes?
no updates in nearly two years
"refer-response"?
does not apply to provisional responses
Robert Sparks - this is radically different from the rest of the
proposals
some debate about whether this is required or not ("not a working
number", etc.)
talk about this offline, based on use case
- add "callid" and "tag" to Refer-to header?
do we need this in the context of an existing subscription, too?
sticking stuff in a Refer-to-URI doesn't work for non-SIP URIs
- Adam - why are all these items in one draft? we have consensus on
some, not others, and maturity levels are very different
can we prioritize them? first two items?
move callid/tag to app interactions? Jonathan is fine (plus
co-authors)
- Robert - sending withour norefersub requires GRUU? Rohan thinks so
- Adam - even first two proposals aren't same maturity level (group
hummm agrees)

- Jonathan - remote CC isn't an appropriate use of SIP - "loosely
coupled but controlled" is a contradiction in terms
- Eric- lots of problems with dorking with a state machine - how can
this work robustly?
- Adam - this is MBUS all over again - MBUS author agrees (Jorg Ott)
- Lots of remote control functions
- People aren't getting escaping right - well, maybe they should fix
escaping
- Jonathan - this is proof that you're using the wrong tool - Refer
was never designed for this.


1425 Exploders
Gonzalo Camarillo
draft-camarillo-sipping-exploders-02.txt
draft-camarillo-sipping-uri-list-01.txt
draft-camarillo-sipping-adhoc-management-00.txt
draft-camarillo-sipping-transac-package-00.txt
draft-hiller-uri-list-index-00.txt
draft-levin-simple-adhoc-list-01.txt

- OMA very interested in these extensions for push-to-talk
conferencing (develop here or they develop their own)
- do we need to support Reply-to-all?
- need a framework showing how all this fits together?
- need to get XCAP URI from server to client
- list management using SIP does not work - remove from draft
- need to deliver or the world will converge
- Allison - Ted Hardie mid-course review needed?
- Cullen - I like this, but the people who care about this can't agree
on the requirements - too early for a WG document
- this is a general mechanism, not a conferencing thing