Notes 每 7/28/2011 MIF Meeting 每 Agenda bashing begins Chair discusses agenda. - Updates to WG items are discussed. 1) ieft-mif-problem-statement-15 has some modifications 2) ietf-mif-current-practices has some minor editorial modifications Point of modifications was the clarify that IETF isn＊t about market analysis. Comment on this draft from Jari Arkko It seems like it＊s a good thing to discuss the different OS behaviors, especially between mobile and desktop. Chair will take a look in response. Chair queried if that was acceptable, no objections raised. 3) Chair describes status of MIF analysis. Last volunteers feel back. Question/Query: Should we drop this? Chair queries for objections if this should be dropped. No objection, chair will work with Jari to drop this. - DNS Server Selection Chair describes updates. This draft is relatively stable. Some minor changes. Chair queries if we＊re ready for WGLC. Ted Lemon with question: He has a draft with some related material, that should be sorted out. Chair: This can be resolved in WGLC. Chair will email WGLC notice to the mailing list. Dave Thaler, Jouni and Gaetan committed to review the draft. - DHCPv6 Route Options Update Tomasz M presents updates to the draft. Basically read through the deck. Draft was discussed at DHCP WG. Discusses updates to ensure that clients don＊t route into black holes. Next version will be provided in 2 weeks. Request is that draft moves into WGLC after next version. Alex Petrescu: This draft might be incompatible with the standard way that default routes are populated. RAs include numerous flags which aren＊t included in the DHCP options being designed here. Alex believes this could be done in a much more full way, and that the mix of RAs and this draft might cause incompatibilities. Chair responds: Please email these concerns. Wojciech Dec joins the debate. There is some back and forth on how the default route is plumbed. How does this draft interop with neighbor discovery? Dave Thaler joins the debate. He comments (paraphrased) ※there is no difference between the information stored for the default route and other routes in host implementations that are RFC compliant.§ - MIF API Discussion Ted Lemon works through his deck. Dave Thaler commentsㄩ This isn＊t really an application API. There are two things that will use this API, the thing that picks what interface you use, and the thing that does the DNS resolution. Chair comments and agrees. Ted also agrees, this is a point he needs to improve upon through the draft/deck/materials. Sri(Cisco): Question about what notifications should happen - what scope of notifications? What if the wireless router is flapping. Ted: In general the API should inform applications when a provisioning domain transitions. Keith Moore: Question about the problems, he is encouraged by the idea. Advice from crowd is he should take a look at the draft (very friendly). Chair and Dave Thaler provide comments Questions for the draft 每 ※Are we providing an API for DNS client, for applications, what layer is this for?§ Ted to take this feedback and specify it in the next drop. Chair polls, who feels good/bad about this draft＊s direction? Most people feel good. Some people feel like more information would be nice 2 objections. Chair requests: ※Please take you objections to the mic. Alex Petrescu and Simon had some objections 每 predominantly that we need much more definition on what API really means in this context. Some debate here. Keith Moore comes out and comments that this doesn＊t really buy me much. Lots of debate. Ted is going to rewrite the draft based on all the feedback. 每 Andrew presents Happy Eyeballs extensions Dave Thaler: Modern applications aren＊t posix-compliant and use many models that match to various abstract APIs. (connectbyname, connectbylist, httpwebrequest, etc). Ted Lemon: Does it really mean for applications to ※connectbyname§? General murmur (Dave/Andrew/audience) 每 yes. Dave, Chair and Yuri and join the debate. Concerns about the security of these APIs, is this really the path we want? Dave: we should rationalize this API with the RFC 3484 API from 6man and make sure both make sense. Ted and Chair jump on this bandwagon. Suggest we build a rational API story for low, mid, and high level APIs for this. 每 Application Test Results in MIF Environment Xiaohong - Walks through deck. Dave Thaler: FYI that the Windows 7 behavior is the same as Windows Vista. Concerning the behavior that Windows waits for a non-nxdomain result to A query before trying AAAA query, this was because people told us they were concerned about Windows killing the DNS infrastructure. Sri (Cisco) 每 Asks if we can increase scope to include multiple addresses and duplicate addresses on multiple interfaces. Chair 每 Suggest that writers look at Ted＊s deck for ideas. Jari 每 Curious about adding literals to the testing. Chair: This is an evolving and living document as the writers add more tests.