PCM Development/Palladio Concall/Minutes 20210112

Aus SDQ-Wiki

Agenda

  • JIRA issues status Dev Board
  • Status PCM 5.0 Release
  • Testing Day (Fr. 15.01.)
  • Clarification of Subsystem semantics
    • Origin: Inconsistent behavior of test case 4 & 5 in SIMUCOM-50 and SIMULIZAR-106
      • Subsystem instantiated twice and allocated directly --> Simulation shall work but produce empty result
      • Subsystem instantiated twice and allocated in a white-box manner --> Simulation shall raise an error
    • Both Subsystem and CompositeComponent are ComposedStructures located in the Repository
      • CompositeComponent can be instantiated multiple times and allocated in a black-box manner
      • Subsystem can be instantiated multiple times (is that sensible?) and allocated
        • in a black-box manner (AllocCtx for Assembly of SubSystem)
        • in a white-box manner (AllocCtxs for each Assembly contained in Subsystem)
      • Black-box allocation of Subsystem is equivalent to use of CompositeComponent
      • Problem: White-box allocation of Subsystem conflicts with multiple Assemblies of Subsystem
    • Suggestion:
      • Short-term Fix:
        • Prevent multiple instantiation of Subsystem (Constraint on ComposedStructure)
        • Reintroduce model validation at the start of SimuLizar run
        • Remove inconsistent test cases and replace by a unit test of model constraint
      • Long-term Fix: Extend allocation context to support "scoping" as part of upcoming meta-model review.
  • SoMoX Status Report

Attendees

  • Anne
  • Floriment
  • Larissa
  • Manar
  • Martina
  • Max
  • Maximilian
  • Sebastian
  • Snigdha
  • Stephan
  • Yves

Release Status

  • StoEx editor, modal dialog
    • working solution exists
    • review in progress
    • finished soon, most probably
  • PALLADIO-451 (Sandro) -> no status
  • PALLADIO-452 (Sebastian) -> still open
  • SIMUCOM-104 (Martina) -> pull request created
  • SIMULIZAR-133 (Martina) -> to be discussed
  • SIMULIZAR-118 (Sebastian) -> in progress, will be done this week
  • SIMULIZAR-133 (Sebastian) -> will be fixed together with SIMULIZAR-118
  • SIMULIZAR-127 (Sebastian) -> in progress
  • SIMULIZAR-132 (Sebastian) -> not started yet
  • summary: too much open points to do testing day and release
    • wait for another two weeks til we can do the testing day / release
    • two weeks should be sufficient to get the tasks done
    • reshedule testing day to Friday, 29.01. 9-12am in morning

Subsystem Semantics

  • SIMUCOM-50 / SIMULIZAR-106
  • Allocation of subsystems
  • background
    • composited components and subsystems are composed structures located in repository
    • difference in allocation / instantiation
      • composite component: multiple instances possible, allocate each instance
      • subsystem: multiple instance possible, whitebox allocation (allocate itself or all of its assembly contexts)
    • problem: multiple instances of subsystems are allowed as of now but this does not work with whitebox allocation
      • allocation of contained assemblies not unambiguous anymore
  • long term fix: extend allocation context to specify a scope
    • nested assembly allocated in context of outer assembly
    • similar to simulation
    • would be metamodel refactoring
  • short term fix (for release):
    • prevent multiple instantiations of subsystems by constraint
    • add model validation as task to simulizar run
    • replace inconsistent manual test cases with automated testing of model constraints
      • strange description of expected behavior (if model is valid in some specific way, simulation works in other situations it does not)
      • remove test 4 / 5 -> define automated tests
  • subsystems are not necessarily meant to be singletons, so long term solution is useful and correct
  • removing subsystems completely would result in loss of whitebox allocation of composed structures
  • agreement on suggested solutions
    • do short term fix for release
    • create ticket for long term solution

SoMoX Status

  • build migration not finished yet
    • migration in process, many bundles migrated
    • a decent amount of bundles might be deletable
  • gast2seff: some code has been removed and compiles now
  • simplemodelanalyzer removed, no replacement yet
  • UI not working yet and not useful without features of Klaus