ChessFidget
You call that chess?
1
The itch
• Like to play chess
• Dislike losing so much
• Too much work to get good
- (For me it would be a dangerous rabbit hole.)
2
The scratch
• Write software to address personal shortcomings
• ChessFidget: a chess app I (currently) always beat
3
Currently looks like this
In lieu of a demo, I'll post code and (Mac) binaries
4
Dev goals/guidelines
• Design for dopamine, not skill
- E.g. no move history
• Computer mostly plays worse chess than I do
- Currently random moves
- That will get boring; see above re: dopamine
• META: Exercise in Swift and reasoning about code
- NOT an exercise in chess programming or AI
5
😬:Agonizing re: modeling
• What are my objects? (Board, Piece, Move, ...)
• In how many places do "business rules" live?
- E.g. MoveValidator and MoveGenerator
- OTOH, might help with mutual unit-testing
• Sometimes the problem was a Swift difficulty
- Extending Grid<T>, T is Piece? vs. T is Bool
• Sometimes the answer was a Swift feature
- Type constraints?
6
"Move" as struct+moveType
enum MoveType {
case resignation
case plainMove
case pawnTwoSquares
case captureEnPassant
case pawnPromotion(type: PromotionType)
case castleKingSide
case castleQueenSide
}
struct Move {
let start: Square
let end: Square
let type: MoveType
}
7
Works fine
awaitingHumanMove
Invalid move e1-g1:
cannotCastleBecauseKingOrRookHasMoved
d1-d2 (plainMove) played by White (Human)
awaitingComputerMove
b8-d7 (plainMove) played by Black (Computer)
awaitingHumanMove
e1-c1 (castleQueenSide) played by White (Human)
awaitingComputerMove
d7-b6 (plainMove) played by Black (Computer)
awaitingHumanMove
8
"Move" as class hierarchy
protocol MoveProtocol {
func updateBoard(_ board: Board)
}
/* The classes would be something like:
Move: MoveProtocol
Resignation
MoveWithStart
PawnTwoSquares
CastleKingSide
CastleQueenSide
MoveWithStartAndEnd
PlainMove
CaptureEnPassant
PawnPromotion */
9
😬:Swift overwhelm
• Building my mental model, esp. re: type system
• Finally past the squirmy phase! 😬➞😊
- New info now gets added to mental framework
• Maybe pairing would have helped?
10
Thoughts about learning
• Know thy brain
- Came to terms with my jumpy learning
- Shlemiel the painter (stay tuned for blog post)
• Be aware when/how reach out to people
11
Other thoughts
• Wow do I suck at chess
• Does writing this app mean I'm a bully at heart?
• Am I weird for sensing uncanny illusion of intent?
- Feels like it's capturing back "on purpose"
- Feels intentional when it checks me
- This despite knowing there is *zero* AI
- WTF it played Fool's Mate
12
Actual early test (Computer=Black)
(manually recreated; wish I'd taken a screenshot)
13
Next (current thinking)
• Find a third-party library, add (weak) AI
• Distinguish between stalemate and checkmate
• Add keyboard input
- Might help me learn algebraic notation
• Add multi-game play
• Move on to other project ideas
14

ChessFidget - You call that chess?

  • 1.
  • 2.
    The itch • Liketo play chess • Dislike losing so much • Too much work to get good - (For me it would be a dangerous rabbit hole.) 2
  • 3.
    The scratch • Writesoftware to address personal shortcomings • ChessFidget: a chess app I (currently) always beat 3
  • 4.
    Currently looks likethis In lieu of a demo, I'll post code and (Mac) binaries 4
  • 5.
    Dev goals/guidelines • Designfor dopamine, not skill - E.g. no move history • Computer mostly plays worse chess than I do - Currently random moves - That will get boring; see above re: dopamine • META: Exercise in Swift and reasoning about code - NOT an exercise in chess programming or AI 5
  • 6.
    😬:Agonizing re: modeling •What are my objects? (Board, Piece, Move, ...) • In how many places do "business rules" live? - E.g. MoveValidator and MoveGenerator - OTOH, might help with mutual unit-testing • Sometimes the problem was a Swift difficulty - Extending Grid<T>, T is Piece? vs. T is Bool • Sometimes the answer was a Swift feature - Type constraints? 6
  • 7.
    "Move" as struct+moveType enumMoveType { case resignation case plainMove case pawnTwoSquares case captureEnPassant case pawnPromotion(type: PromotionType) case castleKingSide case castleQueenSide } struct Move { let start: Square let end: Square let type: MoveType } 7
  • 8.
    Works fine awaitingHumanMove Invalid movee1-g1: cannotCastleBecauseKingOrRookHasMoved d1-d2 (plainMove) played by White (Human) awaitingComputerMove b8-d7 (plainMove) played by Black (Computer) awaitingHumanMove e1-c1 (castleQueenSide) played by White (Human) awaitingComputerMove d7-b6 (plainMove) played by Black (Computer) awaitingHumanMove 8
  • 9.
    "Move" as classhierarchy protocol MoveProtocol { func updateBoard(_ board: Board) } /* The classes would be something like: Move: MoveProtocol Resignation MoveWithStart PawnTwoSquares CastleKingSide CastleQueenSide MoveWithStartAndEnd PlainMove CaptureEnPassant PawnPromotion */ 9
  • 10.
    😬:Swift overwhelm • Buildingmy mental model, esp. re: type system • Finally past the squirmy phase! 😬➞😊 - New info now gets added to mental framework • Maybe pairing would have helped? 10
  • 11.
    Thoughts about learning •Know thy brain - Came to terms with my jumpy learning - Shlemiel the painter (stay tuned for blog post) • Be aware when/how reach out to people 11
  • 12.
    Other thoughts • Wowdo I suck at chess • Does writing this app mean I'm a bully at heart? • Am I weird for sensing uncanny illusion of intent? - Feels like it's capturing back "on purpose" - Feels intentional when it checks me - This despite knowing there is *zero* AI - WTF it played Fool's Mate 12
  • 13.
    Actual early test(Computer=Black) (manually recreated; wish I'd taken a screenshot) 13
  • 14.
    Next (current thinking) •Find a third-party library, add (weak) AI • Distinguish between stalemate and checkmate • Add keyboard input - Might help me learn algebraic notation • Add multi-game play • Move on to other project ideas 14