Notes of the METMOD WG Session at the IETF 76 (Hiroshima) 2009-11-10. Thanks to note takers: Bernd Linowski, Balazs Lengyel Start; Agenda bashing. No changes to proposed agenda. See agenda at http://www.ietf.org/proceedings/09nov/agenda/netmod.txt Working group status report (David Partain). - David Partain: stated that we're late. - David Partain: A great deal of discussion was spent on small issues. - David Partain: Milestones update not planned yet. Plan is to deliver to IESG. Milestones are updated afterwards Common YANG data types. David Partain on behalf of Juergen Schoenwaelder. Slides presented: http://www.ietf.org/proceedings/09nov/slides/netmod-1.pdf - David Partain: Juergen states that he has problems to get people reviewing the document, which is an issue. - 'real' data type support options were discussed. - Wes Hardaker: Is there a use case? - Martin Bjorklund: do nothing - Wes Hardaker: No use case. We already decided not to do anything. - Balazs Lengyel: Do nothing - David Reid: Do nothing - David Kessens: Do nothing - David Partain: against option d) - The strong sense in the room was to do nothing about 'real' data type now (option c) - David Partain: Do last call on types doc immediately after the main yang spec. YANG Status. David Partain on behalf of Martin Bjorklund Slides presented: http://www.ietf.org/proceedings/09nov/slides/netmod-5.pdf - 5 people have read the latest. - David Partain: long discussion about changing a default in the WG mailing list. Discussion resulted in no major changes. - The issue of "system-creatable" was presented. 8 people in the room supported option 2. (Do nothing but write in the document that we know it's an issue but we're still not doing anything.) - The issue of changing a default was presented and discussed. 08 says no update of default. - Balazs Lengyel: Don't change defaults, adding is OK, although netconf monitoring allows finding out the import/include revisions - Balazs Lengyel: vote for option 2 as _changing_ a default could lead to disasters, but _adding_ a default is OK - Wes Hardaker: In SNMP we allowed it, propose 2. This is the last point where we can allow this. - Poll in the room: option 1) no votes 2) 7 + 1 one remote 3) no votes So option 2 (Allow an update to add a default...) was considered to be rough consensus. - David Partain:: asked for review of version 09 as soon as Martin has provided it. Mapping YANG to DSL, Ladislav Lhotka. Slides presented: http://www.ietf.org/proceedings/09nov/slides/netmod-3.pdf - 3 persons indicated that they have read the latest version. - Element order was discussed (Slide 3) Ladislav Lhotka: agreed to make a proposal for a statement for YANG-09 about the order of children of a case and send it to the ML. (second bullet) - David Partain: About slide 3: What exactly do you want in YANG-08? - Ladislav Lhotka: Correct the XML representation text for the case statement. - David Partain: please propose text. - David Partain: What will be the rules for including the documentation info? - Ladislav Lhotka: Documentation statements are readable within the YANG statements. The exact YANG statements might not always be mapped. So there might not be a good place for the documentation statements. - Ladislav Lhotka: the way how documentation statement are handled is left to the mapping implementation. Exclude them as DSDL will be used by tools - David Partain: This is a big change - Ladislav asked the WG to review the schemas in appendix A. - Ladislav Lhotka: introduction of a new namespace URI for XPath extension functions - David Partain: What are the up- and downsides - Ladislav Lhotka: One more namespace URI. - Martin Bjorklund: What about effects on XPath functions? - Ladislav Lhotka: There should be none (?). - Martin Bjorklund: So why is the namespace used? - Ladislav Lhotka: Its usual for xpath extensions - No more comments and question in the room after final conclusion slide. - David Partain stated that external review is needed. Great there is running code used by real customers. Silence is not acceptable to move the document forward. YANG Usage Guidelines Status. Andy Bierman via remote audio. Slides presented: http://www.ietf.org/proceedings/09nov/slides/netmod-2.pdf - 5 people in the room have read the document. - Andy Bierman: are more guidelines are needed? If so, the should be posted to the ML. - David Partain encouraged external review. - Martin Bjorklund (remotely): guideline concerning import with revision. - No more comments or questions in the room at the end of the presentation. NETMOD Architecture Document Status Update by David Partain (no slides) - David asked Phil Schafer (who wasn't present) to update the document with the comments from Juergen Schoenwaelder. - Dan Romascanu: concerned about having a NETMOD architecture document without having (first) a NETCONF architecture document. Did the draft exceed its intended scope, and is it enough? - David Partain: Yes the doc is relatively content free. It would be useful to have a complete netmod/netconf/data model draft. It should be done in Netconf WG. Describe the whole NETCONF picture. This doc will be overtaken by events, but thats not a problem. Complex Typed and Typed Instance Identifiers, Bernd Linowski - Slides presented: http://www.ietf.org/proceedings/09nov/slides/netmod-4.ppt - David Partain: This is not a WG chartered item. It may come only after the chartered items. - David Partain: Which features can be expressed with current YANG features? - Bernd Linowski: Specifying recursive structures with arbitrary/unlimited depth is not possible. Substitution of derived type instance requires explicit model update process. - Balazs Lengyel: Many features can be emulated with current YANG, but it is also import how understandable the result will be be - David Partain: The pyang plugin finished or in development? - Bernd Linowski: Working on it. Todays pyang plugin does not allow full implementation. Working on it with Martin. - David Partain: Please speak with Martin to implement pyang - Bernd Linowski: It is on its way - Bernd Linowski: Inviting interest and co-authors. How to proceed? - David Partain: If aiming for the standard, netmod is the correct WG. This will not happen before YANG is done. After that we can rechartered and maybe include it. Propose a charter update. - David Partain asked how many people have read the document. 7 hands were shown. Dan stated that he would support the document as a contributor because of the relevance of the draft to the mapping between TM Forum and IETF models in the future. - Mehmet Ersue: Earlier Sharon and Alex Chlem, and partially Mark Scott were also interested. Current charter says this is a future problem. - Dan Romascanu: The context was different at that time. - Mehmet Ersue: A yang extension is not a core yang feature. TMForum and ITU is interested in YANG to support inheritance as they use UML a lot. - Bernd Linowski: TMF SID OAM model could be mapped into YANG easily this way. - Mehmet Ersue: TMF would adopt IETF stuff if they could have SID YANG interoperability. - Bernd Linowski: Unlimited containment is impossible in YANG - Balazs Lengyel and Mehmet Ersue: Mapping to YANG possible, the real question is is it usable, is it easy, as we see today it is not. Concluding words - David Partain: - Martin will create YANG-09 quickly. Afters that it will go to WGLC. - As WGLC will be serialized, all other NETMOD documents will be filed for WGLC afterwards. - Balazs Lengyel: Ericsson will propose a draft for actions/RPCs in lists No comments or questions for the open microphone Meeting closed