Engine PR 8.4: king+pawn perft validation #71

New issue

Closed

opened 2026-05-16 16:04:24 +00:00 by root · 3 comments

root commented

2026-05-16 16:04:24 +00:00

(Migrated from gitlab.core-project.space)

Imported from GitHub amaurycoudr/chess-project#71, opened by @amaurycoudr on 2026-05-16 16:04 UTC.
Closed on GitHub on 2026-05-16 21:37 UTC as completed.

Parent

#67

What to build

Prove pawn move generation and legality correct by perft. The reference data already exists: packages/engine/src/perft-fixtures.ts exports PAWN_PERFT_CASES, six positions with Stockfish 18-verified depth/node tables, each isolating a pawn mechanic (double pushes, en passant from the root FEN, en passant created mid-search, push-promotion, capture-promotion, pawns + knight together). This slice consumes that data — it does not pick new positions or re-run Stockfish.

Extend perft.test.ts with a describe block that walks PAWN_PERFT_CASES, mirroring the existing king/knight blocks: each case is parsed from its fen via the file's positionFromFen helper, then driven through the perft helper with an it.each over counts. Choose a per-case depth ceiling that keeps the suite fast (the existing king/knight perft tops out around 90k nodes — match that order of magnitude; the deeper rows in the fixture are reference values, not all meant to run).

Implements step 8 (final part) of docs/adr/0004-engine-slice-plan-and-king-first-piece-ordering.md.

PAWN_PERFT_CASES is deliberately kept out of the running test until pawns exist (see the note in perft-fixtures.ts); wiring it in is part of this slice. The node counts are valid against our engine because every position uses only kings, pawns, knights, and — after promotion — sliders, all of which the engine implements by step 8; no position carries castling rights, the one rule still missing.

Acceptance criteria

perft.test.ts consumes PAWN_PERFT_CASES from perft-fixtures.ts — no hand-picked positions or counts duplicated into the test
Each case is parsed from its fen via the existing positionFromFen helper and validated with the existing perft helper
Every fixture position is exercised; the depth ceiling per case is chosen so the suite stays in the ~100k-node-per-case range, consistent with the existing king+knight perft
The run collectively covers double pushes, en passant (root-FEN and mid-search), push-promotion, and capture-promotion — guaranteed by exercising all six cases
Tests follow the existing perft.test.ts structure (describe.each per group, it.each depth table)
pnpm lint, pnpm typecheck, pnpm test, pnpm build all green from the repo root

Blocked by

Blocked by #69 and #70

> _Imported from GitHub [amaurycoudr/chess-project#71](https://github.com/amaurycoudr/chess-project/issues/71), opened by [@amaurycoudr](https://github.com/amaurycoudr) on 2026-05-16 16:04 UTC._ > _Closed on GitHub on 2026-05-16 21:37 UTC as completed._  --- ## Parent #67 ## What to build Prove pawn move generation and legality correct by perft. The reference data already exists: `packages/engine/src/perft-fixtures.ts` exports `PAWN_PERFT_CASES`, six positions with Stockfish 18-verified depth/node tables, each isolating a pawn mechanic (double pushes, en passant from the root FEN, en passant created mid-search, push-promotion, capture-promotion, pawns + knight together). This slice consumes that data — it does **not** pick new positions or re-run Stockfish. Extend `perft.test.ts` with a `describe` block that walks `PAWN_PERFT_CASES`, mirroring the existing king/knight blocks: each case is parsed from its `fen` via the file's `positionFromFen` helper, then driven through the `perft` helper with an `it.each` over `counts`. Choose a per-case depth ceiling that keeps the suite fast (the existing king/knight perft tops out around 90k nodes — match that order of magnitude; the deeper rows in the fixture are reference values, not all meant to run). Implements step 8 (final part) of `docs/adr/0004-engine-slice-plan-and-king-first-piece-ordering.md`. `PAWN_PERFT_CASES` is deliberately kept out of the running test until pawns exist (see the note in `perft-fixtures.ts`); wiring it in is part of this slice. The node counts are valid against our engine because every position uses only kings, pawns, knights, and — after promotion — sliders, all of which the engine implements by step 8; no position carries castling rights, the one rule still missing. ## Acceptance criteria - [ ] `perft.test.ts` consumes `PAWN_PERFT_CASES` from `perft-fixtures.ts` — no hand-picked positions or counts duplicated into the test - [ ] Each case is parsed from its `fen` via the existing `positionFromFen` helper and validated with the existing `perft` helper - [ ] Every fixture position is exercised; the depth ceiling per case is chosen so the suite stays in the ~100k-node-per-case range, consistent with the existing king+knight perft - [ ] The run collectively covers double pushes, en passant (root-FEN and mid-search), push-promotion, and capture-promotion — guaranteed by exercising all six cases - [ ] Tests follow the existing `perft.test.ts` structure (`describe.each` per group, `it.each` depth table) - [ ] `pnpm lint`, `pnpm typecheck`, `pnpm test`, `pnpm build` all green from the repo root ## Blocked by - Blocked by #69 and #70

root commented

2026-05-16 17:20:21 +00:00

(Migrated from gitlab.core-project.space)

Imported from a GitHub comment by @amaurycoudr on 2026-05-16 17:20 UTC.

Implemented by Sandcastle on branch sandcastle/sequential-reviewer/1778951767805 — awaiting review. Wires PAWN_PERFT_CASES into perft.test.ts with a 110k-node ceiling, exercising all six pawn positions.

> _Imported from a [GitHub comment](https://github.com/amaurycoudr/chess-project/issues/71#issuecomment-4467580321) by [@amaurycoudr](https://github.com/amaurycoudr) on 2026-05-16 17:20 UTC._  Implemented by Sandcastle on branch `sandcastle/sequential-reviewer/1778951767805` — awaiting review. Wires PAWN_PERFT_CASES into perft.test.ts with a 110k-node ceiling, exercising all six pawn positions.

root commented

2026-05-16 17:26:56 +00:00

(Migrated from gitlab.core-project.space)

Imported from a GitHub comment by @amaurycoudr on 2026-05-16 17:26 UTC.

Promoted to branch feat/engine-pr-8-1-pawns-pushes-captures-attack-mask-legality-wiring-68 (PR #72), not yet merged to main. A dependent slice should git fetch origin feat/engine-pr-8-1-pawns-pushes-captures-attack-mask-legality-wiring-68 and branch off it.

> _Imported from a [GitHub comment](https://github.com/amaurycoudr/chess-project/issues/71#issuecomment-4467607607) by [@amaurycoudr](https://github.com/amaurycoudr) on 2026-05-16 17:26 UTC._  Promoted to branch `feat/engine-pr-8-1-pawns-pushes-captures-attack-mask-legality-wiring-68` (PR #72), not yet merged to main. A dependent slice should `git fetch origin feat/engine-pr-8-1-pawns-pushes-captures-attack-mask-legality-wiring-68` and branch off it.

root (Migrated from gitlab.core-project.space) closed this issue

2026-05-16 21:37:09 +00:00

root commented

2026-10-13 10:20:26 +00:00

(Migrated from gitlab.core-project.space)

mentioned in commit 616b4dd697

mentioned in commit 616b4dd697f3e5839bc5c8c16d6d067c5bd53dd8