MINUTES - GEOPRIV - IETF70 Friday December 5 2007 0900-1130 Summary: * Richard Barnes has taken on the role of GEOPRIV working group secretary * We will try to pull together an interim GEOPRIV meeting in early 2008. Watch the list for details. * -http-location-delivery is close to done. There is work to be done in the Security Considerations, a held URI scheme needs to be added, and the document needs to be scrubbed to make sure normative language is only applied to the protocol and not internal logic at any element. Volunteers for detailed review of this document are asked to send mail to the chair. This document will probably go through WGLC in early January. * There will be a draft from Jon and Ted on possible changes to PIDF-LO to address the concerns around redistribution and routing by intermediates that resurfaced while discussing location conveyance in SIP. * The room expressed consensus to adopt draft-thompson-geopriv-lis-discovery-03 as a working group item. * The room expressed interest in pidf-lo-dynamic. Henning will revise the draft before the Philidelphia meeting. We will ask there (if not before) whether the group wants to take this on as a working group item. * The room expressed consensus to adopt draft-polk-geopriv-dhcp-lbyr-uri-option-02 as a working group item. There was lengthy discussion and remaining concern around security that will need to be worked through. * Henning led an impromtu discussion around the taxonomy of location URLs that resonated well with the room. The concepts will be captured in a working group document (TBD). They point to possible mechanism (indicating what kind of location is being requested) that may need to be added to protocols that obtain location references. * The group asked that the civicaddresses-austria draft be reworked into a "Guidelines for placing addresses into PIDF-LO" information document, using the examples from Austria as motivation for its recommendations. ====================================================================== Raw notes as recorded by Miguel Garcia: GEOPRIV ======= 15m Administrivia Robert The agenda is slightly bashed. 30m http-location-delivery (HELD) Mary presentes the slides James: elements and attributes are not the same in XML. Mary: we don't use it consistently. Params should refer to either elements or attributes. Topic: responseTime type Topic: devices that move after the VPN is up. Should we update the text? Topic: Error code extensibility. Topic: HELD URI Topic: normative document dependencies requireing WG adoption/progresion Jon Peterson: teh question is how the general security issues apply to this location acquisition. Richard Barnes: address the security issues individually in each protocol. Jon Peterson: it is a scheduling thing, and presenting items to the IESG in order. We ought to have a story before proceeding to the IESG. James/Hannes: we need Richard's security document into the WG. Resolution: we need the normative references to the security document. Robert: there is a dependency here. There is a candidate here to fill the missing section. The group has to read the document and then decide if we can agree on it and push. A.I to Robert to send an e-mail to the list encouraging people to read the security document and provide comments. Roger Marshall: the document should not put security constraints to other protocols than HELD. For example: the text says the location SHOULD be accurate. Henning: one thing is the external behavior (MUST and SHOULDS). Then there are operational recommendations (good ideas and things to consider), which should be informative. Jon: the text speaks about the function of the LIS box that constitutes the HELD server. A.P. to Mary to inspect the document and look at this kind of normative/informative statements. Version -04 should be ready by early January. Plan to WGLC at the early part of the year. Reviewers requested. http-location-delivery draft-ietf-geopriv-http-location-delivery-03 Mary Robert introduces the discussion of the recipient= parameter in geolocation. There is a disagreement on what this parameter means. Jon Peterson agreed to put a draft. Jon Peterson: The draft will put an indicator inside PIDF-LO that will designate whether routing is permitted. Kind of binary permission flag. In SIP is hard to indicate who is the target of the request. PIDF-LO was thought around a subs/not mechanism. But push model is not compatible with retransmissions. Idea to make special allowance for push systems to make special permission to the routing based on location. Ted Hardie: we should have a draft in Philadelphia to talk about it. It's gonna be ugly. We either need the permission that Jon talk about or we need to change PIDF-LO and change how it works. Henning: we now have a concrete model of what happens to location conveyance. It would be nice to have a set of use cases that describe a bunch of issues (3PCC, 302 redirect). The use cases should look at the redistribution flag and the routing flag. Henning: there is an issue with the reconciliation of legal aspects and protocol aspects. Some lawyers talk about legal entities as opposed to domain names. Comparing URIs with legal entities is not going to solve the problem. We should explore this discussion and bring people who gave us legal advice years ago. Robert: this topic to the list. LIS Discovery draft-thomson-geopriv-lis-discovery-03 James W Lots of people read the document. Hmmm reveals that the group wants to adopt tihs document as WG item. pidf-lo-dynamic draft-singh-geopriv-pidf-lo-dynamic-02 Henning Henning: idea is to extend the LO with dynamic features. Useful for locatoin tracking, fleet management, safety of personnel. The document contains means to cover speed, acceleration, direction, bearing. Topic: units on the wire (m/s, m2/s). Proposal to use a unique set of units. Endpoints can convert if they wish in the User Interface. Resolution: consensus to use a single unit. Rohan: most people want to express two dimentions. The document needs to say that. Rohan: acceleration is a weird word. Call the attribute 'linear acceleration'. Topic: how to indicate use of dynamic feature PIDF-LO in HELD? Two choices: 1- I get to request Geo and you get me whatever you have 2- You explicitly say give me these extensions and if you can't do that, give me an error 3- Please give me acceleration, but if you can't give me civic. Hennig: Proposal, for now do option 1. Then in the future we can extend HELD in a backwards compatible way to request specific acceleration. Rohan: you need to consider filters. A.P. to remind Henning to consider filters for this document. Agreed to do a new revision before Philadelphia. Then decide there whether to adopt as a WG item or not. dhcp-lbyr-uri-option draft-polk-geopriv-dhcp-lbyr-uri-option-02 James P Jon Peterson: I liked it, we need to do it, this is the right solution James W: in the other two DHCP drafts we agreed... Ted Hardie: rather than talking of harmful URI, enumerate the URIs people can expect. You might see the same URI scheme more than once. Hannes: good document, needs better security considerations. There is a discussion on the security properties of DHCP and the document itself. There is a discussion whether posession of the URI means posession of the location. Several arguments at this respect, not clear conclusion. There seems to be quite a lot of confusion to respect the security implications. Robert: comments that the document goes in the right direction. Comments that more stuff is needed. Lisa: we don't understand the security implications. Randal Gellens suggests signing a URI with a revocable key like LEMONADE. Henning discusses the taxonomy of location URLs Hening discusses the 3+1 different classes of URLs that we may have. Jon Peterson throws the idea of having different DHCP options to request a particular kind of URL location. There seems to be consensus in the room that this is a good analysis and is badly needed. Hannes supports the idea of having the ability to request a specific kind of URL. Henning suggests to deprecate the term LbyR and use instead any of the 4 precise URL types. James W: we had LbyR because this is what the requirement said. Robert: we need to capture this in some sort of document. Robert: coming back to the DHCP draft, do we have a conclusion now? Cullen: there is no open issues about the DHCP document. Jon: Disagree, we need a mechanism to request a specific type of URL. Robert 3 questions. 1- adopting the DHCP draft (James P) 2- not adopting the draft 3- leave it in the state it is now, come back later Consensus on adopting the DHCP draft. To be confirmed on the list. 10m civicaddresses-austria draft-wolf-civicaddresses-austria-00 Karl Hannes and James W suggest to reuse an existing field to write a bunch of parts of the civic address. Reason: it is reversible. Austrian gets to know how to use them, for the rest it is an opaque street. It is suggested to turn the draft into an informational RFC about 'Guidelines and considerations when to write Austrian addresses in PIDF-LO'. Robert: revise this, work with authors, bring it as an individual next time. 10m HELD Dereference draft-winterbottom-geopriv-deref-protocol-00 James W While not many people has read the document, many people plan to read it. It is not possible to pick this one as a WG item due to the lack of people who read it. 20m Uncertainty and Confidence James W draft-thomson-geopriv-uncertainty-00 There are comments in favor of this document. There are comments on the relation of the application of uncertainty and confidence, related to the importance of the data. Robert: - Interim around late January - Richard Barnes will join as a WG secretary