Roadmap: chess.js 1.0.0

(Original URL)

After 13 years of pre-1.0.0 development … it’s time.

This is a working document to track planned features/changes and their statuses for version 1.0.0. Please comment if there’s something you’d like to see included or changed. I appreciate and value your feedback. Check back frequently for updates.

I’ll push a basic typescript implementation to dev branch to use as starting point.

Please wait for the dev branch before submitting PR’s.

dev branch is now active

Proposed 1.0.0 Changes

  • Code

    • Rewrite library in typescript
    • Use camelcase for API and everything else
      • enforce with eslint rule
    • Use exceptions in error states instead of returning null. This will allow the library to provide more detailed error messages for common issues (e.g. bad moves, bad FEN, bad PGN). The functions that throw exceptions are listed in the API Changes section.
    • Switch to using the sloppy FEN/PGN parser by default. The phrase sloppy is kind of passive aggressive, so maybe use the terms permissive and strict.
      • load in permissive mode (the default mode) should let the user load FEN without supplying castling rights, ep square, and move numbers (e.g. chess.load(rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w))
      • load and loadPgn should allow an option to use the strict parser (is this even needed)???
    • moves() should work even if there are no kings on the board.
    • Only set the en passant square when the opponent can legally make an en passant move. (see #294)
    • Add a ‘lan’ (better name or is this ok??) property for long algebraic notation (see #259)
    • Improve FEN validation. It should detect common hand-entered such as missing kings and incorrect castling rights.
    • Use fast-check for property-based testing
  • Docs

    • create a chess.js logo (maybe this is something the community can vote on??)
    • Move documentation to chessjs.org
    • Add an FAQ section to answer common questions
  • Other

    • Github issue template ensuring the reporter know the rules of chess before filing a bug (en passant gets flagged as a bug a few times a year)
    • Prefer stackoverflow for chess.js usage questions, github issues for bugs

API Changes

  • game_over -> isGameOver
  • in_check -> isCheck
  • in_checkmate -> isCheckmate
  • in_draw -> isDraw
  • is_stalemate -> isStalemate
  • in_threefold_repetition -> isThreefoldRepetition
  • insufficient_material -> isInsufficientMaterial
  • load_pgn -> loadPgn
  • set_comment -> setComment
  • get_comment -> getComment
  • get_comments -> getComments
  • delete_comment -> deleteComment
  • delete_comments -> deleteComments
  • validate_fen -> isValidFEN
  • The following functions should throw exceptions in the event of an error
    • Chess constructor
    • load
    • loadPgn

Questions

  • Should acronyms be capitalized in function names (e.g. loadPgn vs loadPGN)? Consider the impact this could have on the .fen() and .pgn() functions. I’m leaning towards no, but it’s up for discussion.
  • Should we adopt the chess.js TypeScript definitions from DefinitelyTyped as a basis for our types?

Future Changes beyond 1.0.0

  • use piece lists for increased move generation performance
  • use a parser generator to parse PGN
  • add support for RAV (recursive annotation variations)
  • add support for null moves
  • chess960 support

Issues for First Time Contributors

  • add a default separator=‘\n’ parameter to the ascii() function

Comments (13)

+1

If there’s a major version bump coming anyway, can I suggest the accessor functions that take no arguments be converted to properties? For example:

fen() {
   ...
}

Could become:

get fen() {
  ...
}
+1

Hi all, I just pushed the new dev branch this morning. I manually copied over all of the tests from the 0.13.2 tag and added a few extras where necessary. However, there does seems to be a discrepancy in the number of tests cases jest is running in 0.13.2 (339 test cases) and dev (237 cases). We should investigate this.

+1

Just began using this repo, but this is exciting! I’ll happily make some contributions once the FIX_ME branch is live.

Improve FEN validation. It should detect common hand-entered such as missing kings and incorrect castling rights.

If missing kings is considered invalid, extra kings should also be considered invalid. kkkkkkkk/8/8/8/8/8/8/KKKKKKKK w KQkq - 0 1 is an example board discussed in issue #289

Another request:

Would it be possible to include a few more properties for the’ history’ function? Currently, I’m adding these to the history property in my wrapper:

  • fen
  • check
  • checkmate

Ah, good catch. The main source of your discrepancy is that the validate_fen.test.ts file ran all cases in a single test. See #333 where I’ve “fixed” this in by running each case individually.

Hi @DevAndrewGeorge. No, I haven’t been able to account for the discrepancy in tests cases. I did omit a few cases that I though were redundant on the dev branch, and also refactored a few others. I expected to be within 10-20 tests cases of v0.13.2, but running your snippet on dev shows we’re missing about 100 cases. (BTW - I think you ran on the wrong branch, be sure you’re on dev when generating a test count)

$ git checkout dev && npm test 2>&1 | grep "Tests:"
Tests:       237 passed, 237 total

@jhlywa, did you ever resolve the test issue? I don’t see as big of a discrepancy between v013.2 and dev as you state:

Andrews-MacBook-Pro:chess.js andrew$ git checkout v0.13.2; npm test 2>&1 | grep "Tests:"
HEAD is now at 426e7ed 0.13.2
Tests:       338 passed, 338 total
Andrews-MacBook-Pro:chess.js andrew$ git checkout master; npm test 2>&1 | grep "Tests:"
Previous HEAD position was 426e7ed 0.13.2
Switched to branch 'master'
Your branch is up to date with 'origin/master'.
Tests:       339 passed, 339 total

Hey @jhlywa, I shall do a deep dive of the code before contributing… Any advice on the same…? I’m new to contributions on github… 😅. Also, would be great if you could share some resources on grammar files. I’ve gone through the grammar file of PGNViewer but hitting a roadblock as far as other examples are considered

@coffeeDev98 I’d be interesting in creating the grammar from scratch. I think it would give us a better understanding of how the PGN parser works, and frankly, I think it would be easier than maintaining the existing regexes.

Hey @jhlywa, Had a doubt… How would you go about with creating the PEG file / Grammar file? Will you be creating one from scratch or referring to an existing one like, PGNViewer…? Looking into the PGN Grammar right now…

The variation code will most likely come later in version 2.0.0 (I refer to it above as RAV - Recursive Annotation Variation). The PGN loading code needs to be rewritten from the ground up. It’s currently a set of hand crafted rules that are a little tedious to maintain and follow. A better approach would be to include the PGN grammar and then use a parser generator to generate the parsing code. Once this is done, adding support for variations should be significantly easier.

Hey @jhlywa , Great to see this is back in development!! Have been using the stable package for our project… Had a request… Is it possible to add variations support…? Or Is it something that you’re planning to add…? Like in @aaronfi 's project chess-es6