Skip to content

Phase 25: Engine Wiring & Integration Sprint

Status: Complete Tests: 152 new (6,477 total) Block: 2 (first phase)

Overview

Wired all post-MVP standalone engines (EW, Space, CBRN, Schools, Commander, Era, Escalation) into ScenarioLoader so a scenario YAML alone produces a fully-connected simulation. Fixed the broken OODA DECIDE path (assessment=None, personality=None). Integrated EW engines into the tick loop. No new engines or mechanics — only wiring.

Sub-phases

25a: ScenarioLoader Auto-Wiring (61 tests)

Modified: simulation/scenario.py, c2/ai/schools/__init__.py

  • Added 5 new config blocks to CampaignScenarioConfig: ew_config, space_config, cbrn_config, school_config, commander_config
  • Added 4 new fields to SimulationContext: commander_engine, eccm_engine, sigint_engine, ew_decoy_engine
  • Extended _create_engines() → _create_optional_engines() with 7 sub-methods:
  • _create_ew_engines() — JammingEngine, ECCMEngine, SIGINTEngine, EWDecoyEngine
  • _create_space_engines() — Full SpaceEngine hierarchy (GPS, ISR, EW, SATCOM, ASAT)
  • _create_cbrn_engines() — CBRNEngine + all sub-engines (dispersal, contamination, protection, casualty, decon, nuclear)
  • _create_school_engines() — SchoolRegistry with all 9 schools via factory
  • _create_commander_engine() — CommanderEngine + profile loading
  • _create_escalation_engines() — All 9 escalation/unconventional engines
  • _create_era_engines() — WW2, WW1, Napoleonic, Ancient engine sets
  • Added create_school() factory function to c2/ai/schools/__init__.py with _SCHOOL_ID_TO_CLASS mapping

25d: CommanderEngine Wiring (22 tests)

Modified: simulation/scenario.py, simulation/battle.py

  • Added _apply_commander_assignments() to ScenarioLoader: side-level defaults first, then per-unit overrides
  • Wired commander_engine.get_ooda_speed_multiplier() into battle loop OODA phase advancement (stacks with school multiplier)

25b: Battle Loop OODA Fix (34 tests)

Modified: simulation/battle.py

  • Added _cached_assessments dict to BattleManager (transient, not checkpointed)
  • OBSERVE phase now uses real morale (MoraleState → 0.0–1.0 mapping) and real supply (StockpileManager.get_supply_state())
  • OBSERVE caches assessment for DECIDE retrieval
  • DECIDE retrieves cached assessment + commander personality, builds real assessment_summary for school adjustments
  • Added _get_unit_morale_level(), _get_unit_supply_level(), _build_assessment_summary() helper methods
  • Opponent modeling now uses real force counts

25c: Tick Loop Integration (34 tests)

Modified: simulation/engine.py, simulation/battle.py

  • Added strict_mode parameter to SimulationEngine (default False)
  • Replaced 6 bare except Exception: pass blocks with logger.error(exc_info=True) + conditional re-raise in strict mode
  • Added _update_ew() method: calls ew_engine.update(dt) and ew_decoy_engine.update(dt) when present
  • Wired insurgency with real military presence (active unit counts per side) and collateral data (from consequence engine)
  • Added MOPP speed factor in _execute_movement(): queries CBRN engine _mopp_levels, applies ProtectionEngine.get_mopp_speed_factor()

Files Changed

Modified (5 source files):

  1. stochastic_warfare/simulation/scenario.py — Config blocks, context fields, engine creation, commander assignments
  2. stochastic_warfare/simulation/battle.py — Assessment cache, real OODA wiring, MOPP speed, commander OODA mult
  3. stochastic_warfare/simulation/engine.py — strict_mode, bare except fix, EW update, insurgency data
  4. stochastic_warfare/c2/ai/schools/__init__.py — SCHOOL_ID_TO_CLASS mapping + create_school() factory
  5. tests/unit/test_phase_19e_integration.py — Added commander_engine=None to mock context

New (4 test files):

  1. tests/unit/test_phase_25a_scenario_wiring.py (61 tests)
  2. tests/unit/test_phase_25b_ooda_fix.py (34 tests)
  3. tests/unit/test_phase_25c_tick_loop.py (34 tests)
  4. tests/unit/test_phase_25d_commander_wiring.py (22 tests)

Key Patterns

  • Null-config = disabled: Every optional engine gated by if config.X_config is not None. Zero cost when disabled.
  • Lazy imports: All engine imports inside factory methods to avoid circular dependencies.
  • Safe attribute access: getattr(ctx, "cbrn_engine", None) in battle.py for backward compatibility with SimpleNamespace test mocks.
  • Assessment cache is transient: Not checkpointed — rebuilt each OBSERVE cycle.
  • Strict mode for debugging: strict_mode=True re-raises all engine errors instead of swallowing them.

Deficits Resolved

Deficit Description
1.1 ScenarioLoader auto-wiring
1.2 CommanderEngine not on SimulationContext
1.3 battle.py assessment=None
1.5 MOPP speed factor never passed
1.6 Era engines not wired
1.7 Bare except in engine.py
2.8 COA weight overrides not called (school adjustments now use real data)
4.13 Insurgency needs real data
5.1 EW engines not wired into tick loop

Known Limitations

  • Air campaign ATO wiring (deficit 1.4): Not addressed — would require air campaign engine to expose an ATO planning interface. Deferred to future phase.
  • C2 effectiveness always 1.0: Assessment c2_effectiveness is still hardcoded; requires comms quality integration.
  • Consequence engine collateral: get_collateral_by_region() may not exist on all ConsequenceEngine implementations — gracefully handled with try/except.
  • Stratagem affinity wiring: Planned in 25b but not implemented. get_stratagem_affinity() is not called during DECIDE phase. Deferred.
  • School_id auto-assignment: Planned in 25d but not implemented. CommanderPersonality.school_id does not auto-assign to SchoolRegistry. Deferred.

Postmortem

Scope: On Target (with 2 silently dropped items)

  • 151 tests planned, 152 delivered (1 extra: checkpoint cache clear test added in postmortem)
  • 9 deficits resolved (1.1, 1.2, 1.3, 1.5, 1.6, 1.7, 2.8, 4.13, 5.1)
  • 1 deficit deferred with documentation (1.4 — ATO wiring)
  • 2 items silently dropped (stratagem affinity, school_id auto-assignment) — now documented above

Quality: High

  • All new methods have type hints. No bare print(). get_logger(__name__) everywhere.
  • No TODO/FIXME/HACK in any modified source file.
  • Null-config=disabled pattern applied consistently across all 7 engine groups.
  • Safe getattr(ctx, ..., None) for backward compat — prevented 12 test regressions.

Integration: Fully Wired (no gaps)

  • create_school() called from _create_school_engines() ✓
  • _update_ew() called in tick loop via _update_environment() ✓
  • _apply_commander_assignments() called in load() step 9 ✓
  • ECCM/SIGINT engines are stateless query engines — no update() needed (not a gap)
  • JammingEngine also stateless — hasattr guard correctly skips tick call
  • Only EWDecoyEngine has time-dependent state and is correctly ticked

Deficits: 1 fixed in postmortem, 2 documented, 0 new unresolved

  • Fixed: _cached_assessments not cleared in set_state() — added 1-line clear + test
  • Documented: Stratagem affinity wiring and school_id auto-assignment added to Known Limitations
  • Pre-existing hardcoded values (target_size=8.5, fuel_rate=10.0, echelon=5, auto_resolve params, insurgency economic_factor=0.5) are all assigned to Phase 26b (Configurable Constants) — not Phase 25 scope

Test Quality Notes

  • Hardcoded count fragility: test_escalation_engine_count uses len(result) == 9, test_school_registry_created uses len(all_schools()) == 9. Will break when new engines/schools are added. Acceptable for now — these are Phase 25 tests verifying Phase 25 wiring of exactly 9 known items.
  • Private attribute access in tests: Tests check engine._config, engine._dispersal, etc. This is necessary for construction-verification tests but creates coupling. Acceptable for wiring tests.
  • All tests fast: No @pytest.mark.slow needed. Full Phase 25 suite runs in <1s.

Performance: No Regression

  • Full suite: 6,477 tests in ~102s (consistent with pre-Phase-25 baseline)
  • Phase 25 tests add <1s total