Systematic audit of 1,191 tests found tests written to pass rather than
verify correctness. Key fixes:
Infrastructure:
- TestClock: fixed timezone from .current to America/New_York (deterministic)
- TestFixtures: added 1.3x road routing factor to match production
- ItineraryTestHelpers: real per-city coordinates instead of hardcoded (40,-80)
Planning tests:
- Added missing Scenario E factory dispatch tests
- Tightened 12 loose assertions (>= 1 → == 8.0, > 0 → range checks)
- Fixed 4 no-op tests that accepted both success and failure
- Fixed wrong repeat-city invariant (was checking same-day, not different-day)
- Fixed tautological assertion in missing-stadium edge case
Services/Domain/Export tests:
- Replaced 4 placeholder tests (#expect(true)) with real assertions
- Fixed tautological assertions in POISearchServiceTests
- Fixed Chicago coordinate in RegionMapSelectorTests (-89 → -87.6553)
- Added sort order verification to ItineraryRowFlatteningTests
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Accessibility buttons split the map into equal-width thirds, intercepting
taps before the coordinate-based logic. Tapping the visual West region
could hit the Central button. Adding .allowsHitTesting(false) lets taps
pass through to MapReader's coordinate detection; VoiceOver still works.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
