Requirements

• python ~ 3.8 (tested in 3.9)
• pygame
• numpy
• the path names are for windows (not normalized)

git clone https://github.com/spintronics/connect_four

python connect_four.py

• split up game logic, ui, state, and ai
• test these units independently
• merge them together and do integration testing

• single source of truth state management
• action based mutation of state
• ui is redrawn only when the state changes
• ui is built up using nested components including (image, circle, rectangle) and layout components such as (expanded, row, column)
• event wrapper for components to imbue them with the ability to trigger actions on hove, click, etc
• logic such as checking for a win, dropping a piece, and ensuring a move is valid is seperate from ui and tested independently
• ai looks ahead to first prevent possible wins for opponent then to make the move that gives it the most possible wins (recursively)
• one of the tests tha the ai my pass is to win 100 out 100 games vs another ai that makes random moves
• one of the features implemented to make ui testing easier was to write each action to a log file. If some error occurs during the program execution, we can replay the actions to get the program into the state in which it broke so we can debug.

• initialize state and register action handlers
• build the component tree with the appropriately bounded event handlers
• an event such as a button click fires the bound action
• the action changes the state it a predictable way (immutably)
• if the new state is different, a redraw is scheduled
• the game loop consumes events every 20ms

• pygame, black, flask, requests
• all of the infastructure to enable remote play was implemented but not finalized due to insufficient testing