dnssd WG meeting minutes ------------------------ Chairs introduction: Chairs presented "Note Well". Chairs identified jabber scribe, etherpad scribe. WG status: Chairs to confirm WG milestones are updated as requested. Document status review - Requirements document published as RFC 7558. Several WG documents to be discussed at this meeting. Tim suggested that Doug Otis merge his draft draft-otis-dnssd-mdns-xlink-06 with Hosnieh Rafiee's draft-rafiee-dnssd-mdns-threatmodel. On Interoperation of Labels Between mDNS and DNS draft-ietf-dnssd-mdns-dns-interop-00 Andrew Sullivan The WG discussed WG last call comments and considered comments in the room as WG last call comments. Author will make minor revisions and the WG will review the revisions before requesting IESG to publish the document. Stuart Cheshire asked what is the distinction between IDNA and punycode. Andrew Sullivan responded that IDNA is the lookup scheme, while punycode is the alogrithm that generates RHS of A-label. Further, IDNA doens't allow all of punycode - spaces, mixed case - and the interoperation of DNS-SD and IDNA is the problem. Andrew pointed out that Kerry Lynn correctly pointed out that this document talks about interoperation of DNS and, more generally, other resolution systems. Doug Otis is concerned with homenet IPv6 architecture, and how ".site" is getting entered into the Internet name resolution. Doug says Dave Thaler convinced Doug that any names like /site will be entered into the special use names convention. Doug also noted that ".home" doesn't necessarily make sense in China. Dave Thaler explained "ambiguous" TLD must coordinate with ICANN policies; could always be ASCII, e.g., if that name is never typed. Andrew asked if the document needs more detail. Dave responded that document is sufficient as written. Stuart Cheshire stipulated that there is agreement that leaf names can be rich text. The question is what to do about resolution further toward the TLD. Where are rich text names allowed and where are A-labels required. Perhaps use Mozilla public suffix list. Andrew responded that making the transition one label below the suffix list match may not work but is likely what dmarc will do. Paul Hoffman observed that the document captures one way to interoperate while Apple may do something different, should the WG wait on the document until Apple publishes its methods. Stuart said that sounded like a good idea. Andrew suggested document might give one method if the resolver has no idea what to do, and some other method if the resolver has external information like the public suffix list; however, he would prefer not to make an explicit reference to the public suffix list. The chairs would be OK with parking the document for some time. Question to WG: should Andrew add text about local knowledge to the document? Paul Hoffman answered "yes". Stuart pointed out the fundamental question is where in the name should spaces, et al., be allowed. Andrew said that we need something else on top of DNS, but now we're stuck with DNS. Joe Abley: leaf end is clearly "spaces allowed" while TLD end is clearly "spaces not allowed". The question is where to change from one resolution method to the other. Don't send UTF-8 queries to the root zone, don't waste local resource looking for IDNA/punycode. DNS Long-Lived Queries draft-ietf-dnssd-push-00 Stuart Cheshire The WG reviewed the technology proposed in the doccument. There are actually two separate proposals: long-lived queries over TCP and push notifications. A proposal to split the document into two pieces was put forward, with the work on long-lived queries perhaps taken up in the dnsop WG. The participants in the room were polled regarding whether the dnssd WG should take up some or all of this work, with unanimous consensus in favor. Dave Thaler commented that the WG should adopt this document. Olafur Gudmundsson pointed out that it is really two different documents: one about long-lived DNS connections and DNS push. Markus Stenberg also supports WG adoption. Petr Spacek asked if TLS is a "MUST". Dave Taht supports the work and asks if any work has started on clients. Stuart responded "no" :-(. Andrew Sullivan has some doubts that the work is entirely in scope, but is overall supportive of the work being done. Paul Hoffman asked for consideration of whether TLS and long-livedness should be tied together. Multicast DNS (mDNS) Threat Model and Security Consideration draft-rafiee-dnssd-mdns-threatmodel-03 Rafiee The WG reviewed the latest revision of the document and provided author additional feedback on the scope of the threats covered in the model. The WG participants at the meeting were polled about whether the work should be done in the dnssd WG (unanimous in favor) and whether the document is ready for WG adoption (unanimous to wait until next revision). Review of implementations of Hybrid Proxy for homenet and enterprise The WG held a brief discussion of existing implementations of the hybrid proxy draft and the hybrid proxy autoconf draft in homenet. The autoconf draft is ready to be sent to the IESG (waiting on publication of the hybrid proxy draft), and the dnssd WG was encouraged publish the hybrid proxy draft as quickly as possible.