Skip to end of metadata
Go to start of metadata

Date

10 July 2018

Attendees

Goals

  • IEEE 802.1Qcx CFM DM (June 25th & July 3rd versions) & G.8052.1 UML
  • Photonic & OTSi model (if time allowed)

Discussion Items

TimeItemWhoNotes
IM-AIEEE 802.1Qcx CFMKam
  Bernd
  • Re-engineered UML models
  • Main changes (due to the change in the source CFM YANG):
    • New classes
      • MaintenanceAssociationGroup
        • Need to check with the CFM YANG author (Marc Holness) whether this new class represents the Maintenance Group concept in the previous version (see 2018-07-03 OT-IM Meeting Notes).
      • CfmOperation
      • LoopbackReply
      • MaintenanceAssociationMepList
    • The RPC in the previous CFM YANG now become the Action (YANG 1.1) UML interface; same structure thought
      • RPC was in a separate section in the CFM YANG; It needs to reference the MEG
      • Now the CfmOperation is within ieee802-dot1q-cfm.yang, i.e., under the root element Cfm and no need to reference MEG anymore
    • Deleted object classes:
      • Loopback and Linktrace
        • Need to check with the CFM YANG author whether these two classes have been subsumed into the input parameters of the transmitLoopback and transmitLinktrace operations of the Action UML interface.
        • Need to check with the CFM YANG author whether these two classes LoopbackReply and LinktraceReply contain the result (i.e., the output parameters) of the transmitLoopback and transmitLinktrace operations.
          • Comment: Andrea suggestion instead "Repy" it is better to call them LoopbackResult and LinktraceResult.
    • New Datatype
      • ManagementAddressGrouping
        • Why "Grouping" at the end of the label, the usage is not consistent across all the datatypes.
  • Identified issues (see the list of issues in the MS Word document "Findings_re-engineering_ieee802-dot1q-cfm@2018-06-25" in  oimt2018.BZ.001.05_PapyrusIeeeCfmOxygenWorkspace_180703.zip)
    • Complexity of the paths from MaintenanceAssociationGroup to MaintenanceDomain and MaintenanceAssociation, in particular the "current()" field in the path.
      • Need to understand the intent of the paths.
      • Why both paths are needed. Isn't one association to MaintenanceAssociation is enough?
        • The distinguish name (DM) of the MaintenanceAssociation should already include the DM of the MaintenanceDomain.
    • Mapping issue for Choice (used for ManagementAddressGrouping) given that Choice has been removed from the UML modeling guidelines and only XOR can be used.
      • Note that the options under the choice is a flat list of the options. There is not heirarchical structure, i.e., no nesting subchoice.
      • Also, how to model the default value/option of the Choice?
    • bit value
      • The ieee802-dot1q-cfm.yang has typedef mep-deffects-type, which has type bits with position and decription for each one of the bits
      • The current UML guideline defines bit-related stereotypes for attribute that has bit type datatype, namely the BITS ENUM in the "encoding" stereotype and the bitsDefinition stereotype.
        • The issues with this approach is that
          • It is not possible to define a separate data type that can be used by many attributes
          • It does not have decription
      • Discussed the issue and proposal solution that Bernd has created in https://github.com/OpenNetworkingFoundation/EAGLE-Open-Model-Profile-and-Tools/issues/296.
        • The discussion supported Bend's proposal. In addtion, supported include if-feature, reference and status in the BitDefinition as optional
        • Decision will be made at the IISOMI UML-YANG Mapping call.
    • Mapping of the MUST statement
      • There are MUST statements in ieee802-dot1q-cfm.yang, such as in the "leaf udp-ipv4-domain", that requires the leaf must have a specific value (and the default has to be that value).
        • In UML modeling, it seems that so far we haven't encounter such needs (i.e., an attribute has to take on a specific value even at the model design stage).
          • Anyway, at the discussion, Bernd noticed that it is possible to use OCL to support the MUST requirement.
    • Systematic YANG
      • In ieee802-dot1q-cfm.yang, the container or lists contain directly the leaf(ref)s. But the UML-YANG mapping tool automatically maps each object class to a grouping and inserts the grouping via the "uses" sub-statement into the Container or list. That means the resulting YANG module from the round trip translation will look different from the source YANG model (which has not those groupings). However, this is not a essential issue, because the groupings only exist in the schema tree and never exist in the data tree anyway.
    • Editorial
  • Kam noted that the above issues will be further discussed at the ITU-T Q14/15 modeling coordination vMeeting on July 16 Monday (https://www.itu.int/net/itu-t/lists/rgmdetails.aspx?id=9198&Group=15), which IEEE & ONF & ect. have been invited to participate.
 ITU-T G.8052.1Kam

Not discussed due to running out of time.

Recap of input from the last call.

  • Given that CFM provides only the highest priority (critical) defects, the G.8052 currentProblemList should be retained.
  • Will provide the rationale of pruning/retaining for the attributes in an additional column or excel.
 Photonic & OTSi ModelNigelNot discussed due to running out of time.

Action Items

  •