Notes on SIP Session 2 at IETF 67
Reported by Chris Boulton


Agenda Bash - Chairs



Discussion on loose Route



- Dean - do we except this as a reasonable direction?

- Long line of discussion on grid etc property

- Rohan/Francois Audet concerned about backwards compatibility with new mechanism.

- Robert Sparks - likes current proposal.

- Cullen - do we want a milestone for this?  Doesn't see backwards compatibility problem.  Favor general solution.

- Jonathan R doesn't think there are any spec backward compatibility problems.

- Andrew Allen - No impact P-CSCF - it should work if compliant to specifications.  

Concerns if SBC removes route header.

- CONCENSUS in the room that we have to do something in this space.

- No conclusion today BUT further work is required.

- Rohan proposes a design team to look at use-cases (for backwards compatibility).

- Jonathan - points out that adopting the draft DOES NOT mean that you are adopting the content - just the basis for work.

- Jonathan - No reference from GRUU and so wont hold it up.

- Rohan - disagrees

- Paul - clearly enhances GRUU and GRUU is definitely useable without it.  Would not like to see GRUU held up.

- Scott Lawrence - Proposal has a lot of good properties regardless of GRUU.

- Andrew Allen - also doesn't want to see GRUU held up by this work.  Do it as a parallel piece of work.

- Jonathan - clear that implementing GRUU does not require this - no normative dependency.

CHAIR DEAN - HUM - clear no need to link - Proposal of adopting as base-line - HUM of acceptance.



Location Conveyance - James Polk



- Outlined changes in -05 version

- James explains to Jon Peterson the motivation for the work.

- Paul K - what are the trust issues?  James - entities can lie - Nothing in SIP to stop that.

- Rohan - questions when using location by value - do you need Geolocation.

- Jon Peterson - What is important - who are which entity inserted?

- James - just the type.

- Jon P - any value in adding inside a PIDF-LO object?

- CHAIR Keith - Are we heading in right direction by adding to a header - no one seems to have a problem with current direction.

- James Polk - runs through remaining OPEN ISSUES.

- Line discussion on the use of Content-Disposition - Miguel Garcia suggests using 'signal'.  Take to the list to discuss.

- James Polk - location error created by this draft - needs additional information and so James has written Geopriv draft for an error code registry.  Geopriv hate the idea.

Suggestion to split SIP/LOST etc and have separate registries.

- CHAIR - Keith - suggest conference call before Christmas to discuss.

- James Polk - walks through his 4 alternatives.



Connected Identity - John Elwell



- Current Status - WGLC - fixed except one issue - rejection of SIP by RFC 4474 verifier.

- John - proposal - NOT send 428 to a mid dialog request.  Abandon dialog if you get a 428/436/437/438 to a mid dialog request.  John outlines the pros and cons.

- Jon Peterson - I like it in general - has problems that maybe its too lax (around 427).

- Jonathan R - question to Jon Peterson - what are you worried about?

- Jon Peterson - worried about who is on the other side - there are cases when I care and cases when I don't.

- John Elwell - proposal - don't make statement about 428 and policy.

- CHAIR Keith - POLL - is the draft heading in the right direction?  Proposal seems to be maybe to add an information header in a mid dialog request rather than rejecting out right.

- CHAIR - Keith - John Elwell to go and update doc.



Identity Coexistence - Jonathan R



- Problem statement - how mechanism work together (P-Asserted-ID and Identity).

- Long Line Discussion on inclusion of transient network in solution - currently breaks and folks are concerned about a partial solution.

- Jonathan R - thinks you have to pick your poison - dont use cryptographically asserted messages.

- Paul K - why remove PAID when they trust each other?

- Jonathan R - wanted to keep PAID as intra domain.

- Cullen - Thinks that logic is wrong - rather than looking for PAID and falling back to Identity - should be looking for Identity and falling back to PAID.

- Jon Peterson make point that people are looking to allow 'man-in-the-middle' attacks  to exsit in transient networks.

- CHAIR Keith/Jonathan R - Obvious interest in this work - Jonathan R suggests another individual revision on the document.  Talk to ADs about scoping work.

- Jon Peterson - Agrees that work is valuable and is supported.

- Christer Holmberg - clarification of problem.

 

Outbound Discovery and High Availability - Jonathan R



- Problem statement.

- Jonathan R - draft is very complicated - needs to be split into 3 drafts

(see slides).

- Step through Discovery etc from the draft.

- Scott Lawrence - don't accommodate clients that don't use DNS.

- Jonathan R - agrees that its a side requirement.

- Scott Lawrence - Need some clarity on UA behavior.  Doesn't think you need additional mechanisms if you use single DNS SRV tree.

- BACK TO THE LIST



Handling large UDP responses in SIP - Scott Lawrence



- Problem Statement - responses get much bigger - causing fragmentation in UDP.

- GOAL is not to make UDP work with large responses - simply to flag in responses.

- Walk through of mechanisms (see slides).

- Draft recommends additional responses to flag that its going on.

- Unknown - concern raised that you can only send 100 provisional response to Non-INVITE request - so what happens for things like OPTIONS/UPDATE.

- Rohan - Non-Invite wants to deprecate provisional for Non-INVITE.  He thinks this all is bad.

- Robert Sparks - has seen a lot of phones can deal with fragmented packets.  Have you considered using SIP Redirection?

- Scott Lawrence - that has not been considered at this point.

- Cullen - things that do UDP should be able to re-assemble.  Wanted to make sure those assertions are correct.

- Jonathan R - How do you 'Go and do TCP?'.

- Next Steps - WG interest?  WG item?

- Scott Lawrence - more thought - take to the list.



XCAP Config - Dan Petrie



- Changes from -00 version.

- Problems identified in Montreal - how does it indicate it doesn't support XCAP.  Dan runs through options (see slides).


- draft-petrie-sip-event-param-err-00 - Requirements presentation (see slides).

- Jonathan R - why did we want it as the same event package?  Can't remember motivation.  Jonathan questions the lack of interest in XCAP notifications.

- Mary - confirms that its an OMA dependency

- CHAIR - ACTION - DEAN to ask OMA about their dependency.

- Paul K - thinks that packetcable uses this draft - CONFIRMED.



Route Construction - Jonathan Rosenberg



- Problem statement (see slides) - confusing.

- UA/Registrar behavior not specified.

- Status - re-write to simplify - new backwards compatibility mechanism + very complex 3XX mechanism replaced by loose Route technique (loose Route draft).

- Open Issues - requires input from IMS vendors.

- Rohan - doesn't like using Service-Route for providing the oubound proxy set.

- Christer Holmberg - very useful for IMS as well.



- CHAIRS Close meeting.