Minutes, SIP Working Group IETF 68

Edited by Dean Willis
Notes recorded by Andrew Allen, John Elwell, Lei Zhu, and Bruce Lowekamp
Raw notes from each recorder are available at:
http://www.softarmor.com/sipwg/meets/ietf68/notes/index.html

Jabber scribing was provided by Spencer Dawkins and occasional others.
Logs are at:
http://www3.ietf.org/meetings/ietf-logs/sip/2007-03-20.html
http://www3.ietf.org/meetings/ietf-logs/sip/2007-03-21.html


Jabber transcribing was provided by Renee Cohen
Logs are at:
http://www3.ietf.org/meetings/ietf-logs/sipscribe/2007-03-20.html
http://www3.ietf.org/meetings/ietf-logs/sipscribe/2007-03-21.html

SIP met for two sessions and a total of three hours at IETF 68 in Prague, Czech Republic.

The initial agenda was given as:

Session 1: Tuesday March 20, 2007 1740-1840 Grand Ballroom

Discussion Lead Topic Reading List Est. Time
Chairs Agenda bash, status, announcements This document 5
Jonathan Rosenberg ICE Option Tag draft-ietf-sip-ice-option-tag-01 15
Jonathan Rosenberg GRUU changes draft-ietf-sip-gruu-12 20
Robert Sparks Process for Revising 3261
Mechanism for Publication
Prioritization of Issues
Which current drafts are bugfixes? 		  20

Session 2: Wednesday, March 21, 2007 1300-1500 Congress II

Discussion Lead Topic Reading List Est. Time
Francois Audet SIPS: Guidelines draft-ietf-sip-sips-02 30
Cullen Jennings Outbound draft-ietf-sip-outbound-08 45
If time permits ...
Marc Petit-Huguenin Fragmentation draft-petithuguenin-sip-outbound-fragmentation-02 15
Jae Woo Lee
Henning Schulzrinne		 SIP URI Service Discovery using DNS-SD draft-lee-sip-dns-sd-uri-00 15
Mayumi Munakata Clarification of Privacy draft-munakata-sip-privacy-clarified-00 15
Thomas Froment Record route management draft-froment-sip-record-route-fix-00 10
Vijay Gurbani
Scott Lawrence		 Domain Certificates draft-gurbani-sip-domain-certs-04 15


Session 1

The chairs presented a Note Well slide, made announcements, and reviewed the agenda.

The agenda was accepted mostly as presented, with the first two topics being reversed. It was also noted that Marc Petit-Huguenin would not be present and that the Outbound Fragmentation topic would be deferred.

Topic: GRUU

Lead by Jonathan Rosenberg (JDR)
Slides presented and included in proceedings

Issue: Allowing changes to URI Parameters

Noted that there might be some URI parameters, such as Loose Route, that could require changes.  JDR proposed to make a change to loosen the requirement as currently stated and review existing URI parameters.

Issue: Notification immediately after Reg-Event produces race condition

Proposed solution is to resolve by stating that when call-id of registration changes, weflush all temp-gruus. 

Discussion of change to require minimum time between reg-event and a new notify followed.

The room was polled on preference for the mass-expiry vs. timing, with no clear consensus and approximately an even division between responders. Following a restatement of the poll by Robert Sparks, the majority of the responders favored the proposed bulk invalidation approach. Note that there weren't many responders, and it is not clear that a large number of implementers yet understood the problem or had an opinion on the solution.

Topic: ICE Option Tag

Led by Jonathan Rosenberg (JDR)
Slides presented and included in proceedings.

There was little discussion of the proposal, with no major disagreement noted. However, there  was a suggestion that the change be coordinated with the SIPING working group.

Topic: Process for Critical Changes to RFC 3261

Led by Robert Sparks
Slides presented and included in proceedings

Discussion centered on what is a "critical change" versus an extension. The room generally greed to the process and model of differentiation as discussed.

Area director Cullen Jennings noted that we should lean toward extensions and not critical updates, as that at some undefined point the number of critical changes might lean towards simply revising (and obsoleting) RFC 3261.

Session 2

Note well slide presented and agenda reviewed. Agenda accepted as modified in Session 1.

Topic: SIPS Guidelines

Led by Francois Audet
Slides presented and includeed in proceedings

Issue: Process and Intended Status

Debate initially centered on the Essential Corrections process, with JDR expressing concerning that the proposed split into an explicative document and a bugfix document would be confusing to the readers. The conclusion of this discussion is that the author will attempt to follow the "essential Corrections process and make the document RFC 4485 compliant in the next revision, and we will then revisit and see if it meets our needs.

Issue: Deprecation of Last Hop Extension

The room showed a clear consensus to deprecate the "Last Hop Exception."

Issue: Upgrading during a Proxy Retargeting

This issue was raised in discussion by JDR. Discussion drew out, with the principle opposition being between JDR and Jon Peterson. Following discussion withou resolution, the chairs asked the discussion to be left unresolved at this time and deferre.

Issue: Future Work for True End to End Security

Discussion continued for some time without producing a consensus to do this work at this time in SIP.  Noted that further work to analyze the threat model is probably needed before the value of a solution can be assessed.


Topic: Outbound

Led by Cullen Jennings
Slides presented and included in proceedings

Issue:  Keepalive for TCP, CRLF vs. Inline STUN

Extended discussion of this issue occurred, with each of several factions represented.

It was noted that the current document dropped out discussion of using tags in Path header to resolve configuration questions when Path is used. This is to be repaired in the next revision of the document.

Following discussion the chairs polled on the question of whether the room felt that the issue needed to be resolved "today" and whether they would be willing to accept either approach based on a rough consensus. There was a clear consensus to do so.

The chairs then polled the room on option 1 (STUN) vs option 2 (CRLF). An approximate 2/3 majority favored CRLF.  The ADs decalred "rough consensus" in favor of CRLF. This consensus is to be re-reviewed on the mailing list.

Issue: When to do Keepalives?

Proposed that there is no longer a need for a keep-tcp parameter, and that the keepalive-time parameter will suffice. There was no objection noted to this proposal.

Issue: Simplifying Instance-ID Types

Proposed that SIPS-dependent option be dropped. No objections noted to proprosal.


Issue: OPTIONS Probing for STUN support

Extended discussion occurred.

Proposed that discussion of using OPTIONS be deleted. No objections were noted (just slight whimpering from one of the chairs).

Issue: Procedures for failures within 120s of REGISTER

Proposed that discussion be deleted. A consensus in favor of the proposal was recorded.

Noted that Outbound will be tested at the next SIPIT, and that WGLC for Outbound is imminent, so we should get any issues raised on the list ASAP.


Topic: DNS-SD/Multicast DNS

Led by Henning Schulzrinne
Slides presented and included in proceedings

Discussion occrred without prticular conclusion. Noted that it would be useful to more clearly describe the problem being solved in the dratf.

Topic: Privacy Clarified

Led by Mayumi Munakata
Slides presented and included in proceedings

Noted that work could be divided into two tasks: 1) Clarifying what RFC 3323 says and 2) Extending the solution.

Consensus noted on completing this work. The chairs are to review the approach on-list and work it into the charter.

Topic: Fixing Record-Route

Led by Thomas Froment
Slides presented and include in proceedings

Proposed that the document is effectively a BCP and should be advanced as such.

The room showed consensus in doing the work, but the process (BCP/Bugfix, etc.) remains open.