=============================================================
NETCONF Data Modeling Language WG (netmod)
18th YANG 1.1 Virtual Interim
Monday, August 24th, 2015, 17:00-18:00 CEST
Minutes Juergen Schoenwaelder
=============================================================

* Participants

  AB = Andy Bierman
  IB = Ignas Bagdonas
  JS = Juergen Schoenwaelder
  LL = Ladislav Lhotka
  MB = Martin Bjorklund

* Review of Goals and Milestones

  The WG charter lists the following milestones:

  Mar 2015 - Submit YANG 1.1 to the IESG
  Apr 2015 - Submit YANG guidelines update to the IESG

  Is the goal to wrap up YANG 1.1 with the feature set we agreed on
  during the last year or is the goal to keep YANG 1.1 a moving target
  for some additional time so that additional features can be
  incorporated? This may also touch on the question what can be done
  with extensions. Whatever the answer is, we should agree on a plan
  how to move forward with the YANG 1.1 effort and then set realistic
  milestones.
  
  LL: If we do not finish YANG 1.1 quickly, we have to remove the
      dependency on YANG 1.1 from other documents.
  AB: If we want to address I2RS or OpenConfig requirements, it might
      take anything between a months or a year.
 
  MB: Can we not make extensions work in YANG 1.1?
  LL: We need negotiation of extensions and attributes.
  MB: But this is a different problem.
  MB: You can design extensions such that they do not badly affect
      clients. The NACM extensions work just fine.
  AB: I prefer to update the language instead of having a core
      language and N extensions of it.
  MB: I prefer to have modularity. Eventually certain extensions may
      become part of the language but revising the language every time
      a new extension is needed is costly.
  LL: Should annotations be part of YANG 1.1?
  MB: Annotations define meta-data not regular data nodes.
  MB: I think we should focus on YANG 1.1 but I do not like to find
      myself in a situation that we have to add many additional
      statements once YANG 1.1 is done.
  MB: We had extensions like actions that worked fine because a client
      not understanding an action would never invoke one. But yes, you
      can define extensions that can break things.
  LL: I think the YANG language should address the extension
      negotiation problem.
  AB: There is conformance to YANG, but also conformance to NACM or
      conformance to EPHEMERAL.
  LL: What if I implement ietf-system but not NACM? Do I still have to
      implement nacm: extension semantics?
  LL: I am concerned about vendors misusing extensions.
  MB: Vendors can always break things.
  
  MB: I think we should finish YANG 1.1 soon as we have it now.
  MB: If we have to do YANG 1.2 afterwards, so be it. It would also be
      nice to avoid this if we can work with extensions.
  AB: I do not mind revising YANG every lets say two years if needed.
  MB: I think we agree that we should finish YANG 1.1 now.
  
  JS: What is a realistic timeline to finish up YANG 1.1?
  AB: What about delivering YANG 1.1 in October?
  MB: Works for me.
  AB: Ideally, the next Internet-Draft should go to WG last call.
  MB: Lets discuss next week whether we do another I-D or not.
  
* Actions
  
  - MB will go though the various comments sent to the WG list and
    identify comments that are easy to address and comments that
    require discussion.
  - JS will change YANG 1.1 milestone to October 2015
  - JS will change the YANG guidelines milestone to December 2015