6.5.30 (Stability and bug fixes)

Added

• Added AI/MCP support, including VeraGrid AI modules, MCP server/client
  plumbing, provider orchestration, retrieval, tools, and the initial
  VeraGridMcp package.
• Added broad FMI/FMUs support: metadata inspection, native staging, runtime
  worker lifecycle, FMI 3 metadata/log-category handling, Integer/Int32 runtime
  support, RMS/EMT FMU integration, and external comparison tooling.
• Added major Dynamic Model Editor capabilities: measurements editor/dialog,
  events page/table with filtering, procedural logic editing, DAE code
  assistance, PDF export, diagram dragging, workspace actions, and improved
  routing tools.
• Added DGS/PowerFactory import coverage for static and dynamic models, direct
  FID-based bindings, RMS VSC/HVDC/DC cable handling, graphical operators,
  dynamic roots, and unbalanced network assets.
• Added investment planning features: N-1/PTDF/LODF candidate generation, time-
  series candidate evaluation, candidate verification controls, and new
  investment types for controllable shunts, static generators, and batteries.
• Added new demos and documentation for investments, contingencies, scripting,
  dynamic templates, procedural logic, LFE, Floquet studies, FMI comparisons, and
  validation workflows.

Fixed

• Fixed dynamic editor ownership, stale connections, variable/block renaming,
  route persistence, deleted-block connection intents, imported diagnostics,
  switch events, and modal/window lifecycle issues.
• Fixed multiple GUI clean-crash paths, including contingencies, toast widgets,
  menu blocking, session threading, fluid/turbine editing, and simulation window
  locking.
• Fixed DGS/PowerFactory import issues around VSC RMS assembly, meter ordinals,
  template rounding, tower lookup/susceptance handling, generator control mode
  parsing, and dynamic graphical bindings.
• Fixed simulation and analysis issues in CPF, nonlinear OPF chunking, RMS
  initialization, native RMS FMI validation, EMT grounding migration, VSC/linear
  power flow behavior, DC/linear-analysis matching, and Pareto plotting.
• Fixed data/model handling issues including profile assignment for enums,
  profile import plotting, saving behavior, DB operations when tables change,
  RAWX file closing, incomplete RAW file handling, and missing DC-bus voltage
  assumptions.
• Fixed investment evaluation issues around redundant variations/diagrams,
  inactive generated investments, upgrade investment behavior, clustering
  results, and candidate grouping.

Improved

• Improved the Dynamic Model Editor structure, block properties, events
  workflow, measurements workflow, routing behavior, catalogue organization, and
  RMS/EMT template handling.
• Improved GUI robustness with self-locking workflows during simulations,
  better cancellation behavior, fewer blocking dialogs, wider hit areas, better
  object selectors, and better table/filter interactions.
• Improved internationalization with large translation updates, more
  translatable UI strings, translated DB names/results, translated diagrams, and
  updated language assets.
• Improved NTC reporting, time-series NTC results, clustering integration, and
  Pmode3 behavior.
• Improved investment objectives so losses, overloads, voltage magnitude/angle
  deviation, CAPEX, and OPEX align with documentation; Pareto plots now focus on
  overloads versus CAPEX.
• Improved documentation, generated API/build docs, dynamic-template docs,
  scripting help, and validation examples.

Performance

• Improved NTC convergence behavior and reported faster Pmode3 operation.
• Improved GUI responsiveness by making Matplotlib plotting non-blocking and
  reducing redundant post-run dialogs.
• Improved dynamic routing and connection-line handling, including optimized
  path handling and better obstacle-aware routing.
• Improved scalable short-circuit work with dynamic evaluation at infinity.
• Improved RMS/FMI initialization and runtime paths through vectorized
  initialization contracts, persisted parameter initialization, and isolated
  runtime workers.
• Improved deterministic numerical behavior by applying stable sorting in
  relevant analysis paths.