Peer to Peer SIP (p2psip) IETF80 Thursday, March 31 Minutes by Peter Musgrave RELOAD base(Cullen) any issues with: -  new DRR text? (no) - configuration refresh? (no) - Configuration Defaults? (no) - Resolving Simultaneous Connects Ekr - need to excise a sentence in the doc to be consistent. - Replica/Topology (no) - Triggered Configurations …rest of slides…(no) Open Issue: How to get a cert with multiple node IDs (proposal has no objections) Open Issue: Signing with multiple node IDs Ekr: IF you have N with M then need to do all N incarnations - but in general will be small numbers. Not a big deal but worth pointing out. Cullen: Agree Roni: With a single node would you still hash? Cullen: Yes. MArc P-H: In my proposal used a new enum. Second enum using this hash, so less implementation issues. Ekr: So not backwards compatibility. But I'm ok with this. Cullen: WIll do the proposal and keep the old one for backwards compatibility. Ekr: So we only send old way if don;t need new way Cullen: We will make this concrete and present a slide at the end of the meeting Identifying Connection Endpoints: [Cullen] This is a design taste issue [Marc] One special problem case, client does not send ATTACH. Suggestion to use PING [Cullen] Bruce's argument against is messaging. [Ekr] Don'r want to design without Bruce [Brian] We do have time. [Cullen] I'll get Bruce and Marc together and we will work it out. Signing Stores [Ekr] If you know finger table and that person is far away and can;t replicate - should reject. But basically harmless. [Cullen] Could be difficult to do in some DHT algorithms [Cullen] Proposal is ok? (yes) Refreshing Finger Table - will fix text Way forward Revised slide for multiple node ID signing… [Brian] I will upload to meeting materials. [Marc P-H] I have a list of other issues…(?) Do we need a type for wireshark… (??) [Gabriel] Question about store exists=false, does it have a lifetime? [Ekr] Treated exactly like other stores. [Brian] Let's have some text on this. We'll look for a new version, then we'll move this along. Direct Response/Relay (Roni Evan) Opinions about DRR/RPR? [Cullen] I prefer these as two drafts. [Roni] Difference is not large… [Cullen] May be cases where you can use DRR when you don't have relays, but still think can have a relay draft but want it to be separable. [Brian] But could we make that clear in one doc? [Roni] So issue is whether we can state support by referencing one RFC. So let's have two docs. (agreed. DRR first, then we'll run through the process on the list before the next meeting) Shared Resources in RELOAD (Gabriel Hege) [Ekr] Whats the access control policy on the access list entries Can only be written by peer which created them. [Brian] We have to many things to do to take this on now Who has read this? (about 6-7 people) [Brian] Is this interesting? [Cullen] Yeah, this is cool. Might be useful. Maybe need use cases? [Cullen] This authorizes whole classes i.e. a whole KIND. Do we need to think about more carefully? Network Mgmt Scenarios (Peng Yongling) Any feedback? [Dan R] When use RELOAD to decide where SNMP notifications go need some method to xfer information gained about session into SNMP target. There is a standard way to configure with the target MIB. [Peng] We are thinking of both directions. Manager needs RELOAD to find node. We may need more than one manager. May be an issue to statically configure in endpoint [DanR] Don't disagree - but there are some existing SNMP ways to have multiple manager dynamically, this is now another method for discovering - but this info needs to be xfered to places where standard SNMP agents would look for it. [Brian] List discussion please. Access Control Policies (Marc P-H) If you are designing a new access policy, I would like to see it. [Ekr] Have you thought about impact of CPU cost on recipients? Do we need a rule which limits number of instructions executed? [Ekr] Is there proposed text for mandatory extensions [M P-H] I have some ideas - I'll send them. [Brian] Do we need to define a sandbox about what ECMAscript can do? [MPH] Agree. BCP (Zhang Yungfei) [Brian (as individual)] Why is there a 20s tear-down - this seems weird (even considering firewalls etc.) [YF] (?didn't get response for notes) Disco (Alexander Knopf) [Brian] Appears people are interested - but we should progress an individual until our backlog clears.