2nd/10_Wiki/Topics/Architecture/Tetris_Project_Retrospective.md

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

id	title	category	status	canonical_id	aliases	duplicate_of	source_trust_level	confidence_score	verification_status	tags	raw_sources	last_reinforced	github_commit	tech_stack
wiki-2026-0508-tetris-project-retrospective	Tetris Project Retrospective	10_Wiki/Topics	verified	self	Tetris Retro Tetris Postmortem	none	B	0.85	applied	retrospective project tetris game-development lessons-learned		2026-05-10	pending	language framework TypeScript Canvas/React

Tetris Project Retrospective

매 한 줄

"매 small game project = 매 architecture concept 의 condensed lab". Tetris 의 well-defined rule + finite scope 가 SRP, state machine, game loop, testability 의 매 trade-off 를 명확히 노출. 매 retrospective 는 매 lesson 의 codification.

매 핵심

매 lesson layers

• Architecture: pure game logic vs render separation.
• State: deterministic state machine — 매 replay 가능.
• Testing: pure logic 의 100% coverage, render 의 visual test.
• Performance: 60fps 의 frame budget 16.67ms.

매 핵심 결정

• Immutable state: 매 board 의 매 frame 의 new copy — 매 undo/replay free.
• Tick-based loop: 매 wall-clock 분리 — deterministic test.
• Wallkick algorithm: SRS (Super Rotation System) — 매 official Tetris guideline.

매 응용

1. 매 게임 logic 의 pure function 추출 — UI swap (Canvas/SVG/Terminal) 자유.
2. 매 deterministic seed RNG — 매 bug reproduction.
3. 매 retrospective template 의 reuse for 다른 small project.

💻 패턴

Pure game state

type Cell = 0 | 'I' | 'O' | 'T' | 'S' | 'Z' | 'J' | 'L';
type Board = ReadonlyArray<ReadonlyArray<Cell>>;

interface GameState {
  readonly board: Board;            // 20 x 10
  readonly active: Piece;
  readonly next: Piece[];
  readonly hold: Piece | null;
  readonly score: number;
  readonly level: number;
  readonly linesCleared: number;
  readonly tick: number;
  readonly status: 'playing' | 'paused' | 'gameover';
}

Reducer-style transitions

type Action =
  | { type: 'MOVE'; dir: -1 | 1 }
  | { type: 'ROTATE'; cw: boolean }
  | { type: 'SOFT_DROP' }
  | { type: 'HARD_DROP' }
  | { type: 'HOLD' }
  | { type: 'TICK' };

export function reduce(state: GameState, action: Action): GameState {
  switch (action.type) {
    case 'MOVE': {
      const moved = movePiece(state.active, action.dir, 0);
      if (collides(state.board, moved)) return state;
      return { ...state, active: moved };
    }
    case 'ROTATE':
      return tryRotate(state, action.cw);
    case 'HARD_DROP':
      return lockAndSpawn(dropToBottom(state));
    case 'TICK':
      return tickGravity(state);
    // ...
  }
}

SRS wallkick

// JLSTZ piece kick offsets (rotation 0 → R)
const KICKS_JLSTZ_0_R: Array<[number, number]> = [
  [0, 0], [-1, 0], [-1, 1], [0, -2], [-1, -2],
];

function tryRotate(state: GameState, cw: boolean): GameState {
  const rotated = rotatePiece(state.active, cw);
  const kicks = lookupKicks(state.active.kind, state.active.rot, cw);
  for (const [dx, dy] of kicks) {
    const candidate = { ...rotated, x: rotated.x + dx, y: rotated.y + dy };
    if (!collides(state.board, candidate)) {
      return { ...state, active: candidate };
    }
  }
  return state;  // rotation rejected
}

Deterministic 7-bag RNG

function* sevenBag(seed: number): Generator<PieceKind> {
  const rng = mulberry32(seed);
  const kinds: PieceKind[] = ['I', 'O', 'T', 'S', 'Z', 'J', 'L'];
  while (true) {
    const bag = [...kinds];
    for (let i = bag.length - 1; i > 0; i--) {
      const j = Math.floor(rng() * (i + 1));
      [bag[i], bag[j]] = [bag[j], bag[i]];
    }
    yield* bag;
  }
}

Game loop with fixed timestep

const TICK_MS = 16.67;
let acc = 0;
let last = performance.now();

function frame(now: number) {
  const dt = now - last;
  last = now;
  acc += dt;
  while (acc >= TICK_MS) {
    state = reduce(state, { type: 'TICK' });
    acc -= TICK_MS;
  }
  render(state);
  if (state.status === 'playing') requestAnimationFrame(frame);
}
requestAnimationFrame(frame);

Pure logic test

it('clears completed line', () => {
  const board = makeBoard([
    /* ... 19 empty rows ... */
    ['I','I','I','I','I','I','I','I','I','I'],  // bottom full
  ]);
  const state = { ...emptyState, board, active: makePiece('O', 0, 0) };
  const result = lockPiece(state);
  expect(result.linesCleared).toBe(1);
  expect(result.score).toBe(100);  // 1 line × level 1
});

매 결정 기준 (lessons)

결정	결과	재사용?
Immutable state	매 replay/undo trivial	✓
SRS wallkick	매 standard 준수 — player familiarity	✓
Canvas render	매 60fps 안정	✓
React render (실험)	매 30fps drop 발생	✗
Pure reducer	매 unit test 의 ease	✓
Web Workers (실험)	매 overhead > benefit (small state)	✗

기본값: 매 small game = pure reducer + Canvas + fixed-timestep loop.

🔗 Graph

• 변형: State Machine
• 응용: Game Loop
• Adjacent: Testability_Architecture · Technical-Architecture · Test-Driven_Development

🤖 LLM 활용

언제: small game architecture 설계 reference, retrospective template, SRS wallkick 구현 question. 언제 X: 매 specific Tetris ranking algorithm (TGM, etc.) 의 deep detail — 매 specialized guide 필요.

❌ 안티패턴 (실수 회고)

• Mutable board: 매 undo/replay 가 fragile — debug 의 nightmare.
• Render in reducer: 매 testability 의 손실 — separate concerns.
• Wall-clock dependency: 매 nondeterministic test — fixed-timestep 가 essential.
• Ad-hoc rotation: SRS 미준수 — 매 player 의 muscle memory 와 conflict.

🧪 검증 / 중복

• Verified (Tetris Guideline 2009; SRS spec; project commit history).
• 신뢰도 B (project-specific retrospective, generalizable patterns).

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — full retro with SRS + reducer + lessons table