Skip to main content
Jump to: navigation, search

VIATRA/DeveloperMeetingMinutes/Meeting20180111

Topics

  • Current status: Bugzilla Gerrit AERI
  • VIATRA 1.7.1
    • Proposed release date: January 31st 2018 (together with 2.0 M2)
  • VIATRA 2.0 status
    • UI test failures in recent builds
    • Open questions:
      • Guava usage
        • Most usage with trivial replacements removed from runtime
        • Question: Complex collection replacement: Multiset, Multimap, Table
        • Question: Utility methods replacement, such as Sets.union, Sets.powerset, etc.
      • Language design: multiple languages vs. pluggable scope dialects
        • Pluggable superlanguage version is much more flexible; would offer many benefits:
          • “Match this pattern over this EMF model, plus this DOORS/etc model, plus this traceability table between them” → easy with hybrid scopes
          • “Template patterns”, restrict certain patterns (both in the incremental and the stateless case) to a few user-supplied starting points (objects or identifiers) at runtime → this can be a use case for a user-supplied (unary) table
        • Pluggable superlanguage may take more initial effort
        • Runtime-level plus language-level composition
          • Solve editor content assist etc. issues…
          • import <<IVQLImportable on classpath>> <<details>>
            • Example importables: epackage, java, user-table etc. ← list should be extensible; also probably deprecate / remove the default case
            • “Details” could be Java package name, EPackage nsURI, user table name/class etc.
            • Give +1000% time budget to Zoltán :)
        • Timeframe? Is it meaningful to release 2.0 with language issues postponed?
      • Feature plans: https://docs.google.com/document/d/125Le7ejjXtLf1Xcz2Opra7Afs6_mm9-ciNXRGDJmmts/edit?usp=sharing

Minutes

  • 1.7.1 release: planned for Jan 31.
  • RCPTT UI test failure
    • Zoli still struggles with finding root cause
  • Guava usage
    • Priority should be to remove from API
    • Internal code can use Guava, if suitable (even in runtime with no Xtext/Xtend dependency)
    • Known issues: Preconditions when using Guava 19 with code compiled with Guava 21
    • Retroactive approach: add specific rules to Sonar analysis
  • Language design
    • No strong motivation for “superlanguage” at this point
    • Merge patternlanguage and patternlanguage-emf languages

Back to the top