Child pages
  • 2020-01-31 IISOMI Meeting Minutes
Skip to end of metadata
Go to start of metadata

Date

Attendees

Info

Agenda

  • Administrative
    • Call plan (schedule, leader, topics)
  • Open Discussion

Discussion items

10 minCall planAll
  •  Leader Andrea Mazzini
  •  Leader Martin Skorupski (ITU-Meeting; Kam & Scott can't join)
  •  Leader Martin Skorupski (ITU-Meeting; Kam & Scott can't join)
  • Skip (MEF-Meeting; Andrea & Karthik & Martin can't join)
  •  Leader Bernd Zeuner(O-RAN-Meeting in parallel; Martin can't join)
  •  Leader Hing-Kam Lam
  •  Leader
10min

Tooling issues 

0 minMapping tool issues found during mapping of G.8052.1

Issues have been found when analyzing the errors that occur during the mapping of the G.8052.1 UML.

Scott Mansfield presents a number of found issues:

  1. Import of items which are originally defined in IEEE (or IETF) YANG modules. Clarified that:
    1. IEEE YANG modules are reverse engineered in UML/Papyrus models
    2. ONF models import these IEEE reverse engineered UML models
    3. ONF models are translated in YANG modules by the UML-YANG tool
    4. IEEE reverse engineered UML models shall not be translated back in YANG by the tool, as the resulting YANG modules will be unavoidably different from IEEE original YANG ones.
    • Conclusion is that Specify stereotype needs an enhancement, i.e. a way to indicate whether the tool shall generate YANG from the specified UML (current behavior) or directly use the specified YANG definition.

  2. According to YANG rules, a leaf cannot start with a number. Found cases where the tool fails to translate e.g. "1" into "one".
  3. Identities, the "identity PROBABLE_CAUSE" was missing.
  4. UML Enumerations, there are two possible translations into YANG:
    1. Enum (when "Is Leaf" option in Papyrus advanced properties is set to true)
    2. Identity (when "Is Leaf" is set to false - which is the default value)
      • In both cases there are translation issues.
  5. CurrentData, defined as abstract class: the generated YANG includes only an empty grouping and not the container, which is necessary to position the class in the tree. After analyzing the related UML, the issue is likely caused by a reference from ThresholdProfile class to CurrentData abstract class ("_associatedCurrentDataList"). Decided to replace this reference with a number of references involving not-abstract subclasses (1dm, Dm, 1Sl etc.) of CurrentData.
  6. Grouping threshold-profile, the tool generates "dm-threshold-list" twice (and also other types).
  7. Leaf current-data-id is defined as object key, but this is in conflict with the "config false". Bernd Zeuner confirms that in UML all object keys are defined as config-false, while it appears that in YANG some keys must be writable. Mentioned that Martin Skorupski has solved this issue with a "must" statement.
  8. Some types in YANG modules are still in upper camel case (e.g. Real, TimeUnit, GeneralizedTime, TimePeriod).
0 minAgenda of next callAll

Call plan for FEb 07:

  • we will see


Action items

  •