Sudoku

Sudoku is literally a constraint satisfaction problem: no duplicates in rows, columns, or 3×3 boxes. The game rules map 1:1 to Directive’s constraint–resolver flow.

1. Facts – Grid state, solution, givens, timer, selection, notes, difficulty
2. Derivations – Conflicts, progress, timer display, same-number highlighting, candidates (auto-tracked, no manual deps)
3. Events – selectCell, inputNumber, toggleNote, requestHint, tick, newGame
4. Constraints – timerExpired (priority 200), detectConflict (100), puzzleSolved (90), hintAvailable (70) – evaluated by priority after every fact change
5. Resolvers – Handle game won/lost, increment error count, reveal hints
6. Effects – Timer warnings at 60s and 30s, game result logging

The constraint cascade is the key insight: when a player types “5”, the grid fact updates, derivations recompute (conflicts, progress, isSolved), then constraints evaluate by priority – detecting conflicts, checking for a win, or firing a hint.

What: A fully playable Sudoku puzzle with multiple difficulties, notes, hints, undo/redo, and a countdown timer.

How: The game is a single Directive module with 14 facts tracking grid/selection/timer state. Derivations auto-compute conflicts, progress, and candidates. Four prioritized constraints cascade on every input to detect conflicts, check for a win, expire the timer, or reveal hints. Resolvers handle the outcomes. Effects fire timer warnings.

Why it works: Sudoku is a constraint satisfaction problem – Directive’s constraint–resolver flow maps 1:1 to the game rules. No imperative state machine needed; declare what must be true and Directive handles the rest.