Agenda bash 5 Chair: Asked who had read record-route-fix-03, just finished WGLC - only one person, so need to assign reviewers so that it gets read by Sunday. Reminder on WGLC of dtls-srtp-framework-02. Companion in AVT also in WGLC. Domain-certs-01 now updates RFC 3261 and is now Standards Track. All to make sure they are happy with this, so any further comments needed very quickly. Publication about to be requested. Locaton-conveyance-10 will have another 1 week WGLC. New milestones. Identity discussion on Tuesday inconclusive, so participants asked to continue on list. Also on the further suggested work item of a document documenting existing identity mechanism in terms of security expectations (on Tuesday's slides but not presented), the chairs intend to open a work item on this. v Mechanisms for UA Initiated Privacy: Mayumi Munakata 25 John: Need to make it clear that Privacy: id still needs to be sent. Martin Thomson: Need to consider impact of location-routing-allowed header field, being discussed in GEOPRIV. Richard Barnes: This is not a problem, because default is not to allow insertion. Nils: Question concerning information added to Record-Route. John: The GRUU will identify the domain name of the user. Mayumi: You need to use a GRUU from third party domain if you want to hide the domain name. The document needs some more text to describe this. Chair: When these issues are fixed, is it ready for WGLC? John: It also needs a bunch of editorial changes, which I have sent to Mayumi, but when those are done it will be about ready. Chair: So WGLC will start shortly after next version becomes available. Robert: Please wait until after SIPIT in October. Chair: OK v Termination of early dialog prior to final response: Christer Holmberg 20 Open issue 1: should it be allowed to send 199 reliably. Proposal to allow sending it reliably, but you don't need to even if you have received Require: 100rel. Robert: So this is asking to violate the provisions of RFC 3262? Does not find the motivation sufficient for breaking RFC 3262, and does not believe this would be part of a future HERFP solution. So if someone asks for 100rel, they have to forego any expectation of receiving 199 when a proxy sends it. Francois: Because 199 is sent by a proxy, it should be treated more like a 100. Can't get a proxy to send 199 reliably. Christer: So issue of whether proxy can send 199 at all if 100rel has been requested. Chair: In summary, we want to preserve requirements of RFC 3262. Robert: If proxy sends 199, it is effectively acting as UAS and subject to UAS rules for 100rel. We can't resolve this now, but I will help off-line. Chair: We need to identify text that does not involve updating RFC 3262. Continue on list. Open issue 2: can a UAS be allowed to send 199. Proposal: Allow UAS to send 199. Robert: This is the right thing, and you can expect more text from me. Want UASs to use it. John: A proxy acts as a UAS when it sends 199, so no issue. Open issue 3: Should 199 contain information concerning the final response that triggered it. Proposal: If the initial request indicates support of sipfrag, recommended to include sipfrag message body in 199. Robert: I have already objected to anything as strong as SHOULD, particularly if motivation is solution to HERFP, because it doesn't solve HERFP. Francois: If we don't do that now, implementations will go on market for 199 and will later have little motivation for doing more for fixing HERFP. Uncomfortable with going forward with 199 while still in dark about how we will solve HERFP. Does not believe the other problems solved by 199 are sufficient motivation for publishing this draft. Chair: If you think that work is required, it requires a new draft. Robert: When we agreed to adopt this as a WI, we explicitly said HERFP was outside scope. Scott: Agrees with Robert. I can help with text. Doesn't prevent any UA putting such information in a sipfrag body. Would help to gain implementation experience. Open issue 4: Do we need an option tag, so that UAC can declare its interest in receiving 199. Proposal: Do not specify an option tag. Cullen: Not sure. I am getting more and more concerned there are networks that won't work unless you have 199. Christer: No. This only allows resources to be released earlier. Cullen: Proxy doesn't maintain state for early dialog. Christer: There are 3GPP uses. Cullen: Re-iterates concern. Chair: Need to review draft to ensure it includes appropriate compatibility mechanisms. Cullen: Concerned with fact we keep hearing discussions around features that were not part of the original use case on which we based the work item - scope creep. That initial one requirement does not need an option tag, so why do we need it? Christer: But needed to avoid unnecessary traffic. Cullen: That is a good answer - I am satisfied. Chair: Conclusion - don't include option tag, but add more discussion on compatibility. v Keepalive Without Outbound: Christer Holmberg 10 Christer: Present 4 use cases. Chair: Little discussion in list. Is there anything broken in what Christer has been saying, such that we shouldn't be doing this in first place. Remi: Is there a requirement in 3GPP? Christer: There is. Remi: Why not use options? Christer: Need a keep-alive between UE and edge proxy. Options is e2e, and also too much overhead. Chair: Lets go through these use cases and see whether they are problems to be solved. Aki: But is the problem already solved by other means? Bruce: Question concerning use for P2PSIP. Jonathan: Uses STUN. Francois: Use case 1 (poor man's outbound) and IPPBX trunking are probably the more interesting use cases, but does not have a strong opinion whether we should do it. Chair: How many think it is absolutely necessary to solve this for use in absence of outbound? About 25. Jonathan: Use case 3 (intermediate-intermediate) is the most useful, and can achieve a much faster detection of a broken connection than options (which is used now). Chair: Out of time, but list discussions should try to identify how many of these use cases we really need to address. v Guidelines for double route recording: Thomas Froment TBD No time. John