Onitama App

Can be played at https://onitama.app/

Roadmap

Things that could be cool to implement, that aren't done yet, and might one day get done

• Show piece that last moved, and where it moved from
• Say when opponent has requested a rematch
• Add chat (maybe)
• Add "how to play"
• Add different difficulty AIs

Local AI

The default Dockerfile will disable server-side AI, and instead compile the AI Agent as web-assembly and run in a JS WebWorker.

An alternate Dockerfile (Dockerfile.remoteai) uses server-side AI which runs the agent code in the same process that serves the game and delivers messages for multiplayer games.

The server-side AI is roughly 30x faster at running Monte Carlo simulations, meaning that the hard bot is notably harder when run server side.
The server side AI is roughly 2x faster, ever since swapping out the RNG used in the Monte Carlo search. Previously it was significantly slower.

To use server-side AI, build the container with

docker build -t onitama:remoteai -f Dockerfile.remoteai .

As of writing this, https://onitama.app/ uses the local AI as it is very light on server resource requirements.

Deployment

Run from GitHub

Pull from GitHub container registry

docker pull ghcr.io/jackadamson/onitama:latest

Run the container

docker run -dp 80:8080 --name onitama --rm ghcr.io/jackadamson/onitama:latest

build Locally

build the container

docker build -t onitama .

Run the container

docker run -dp 80:8080 --name onitama --rm onitama

Development

Requires Rust (nightly) and Node (v14)

1. Install dependencies with yarn install
2. Start the backend with cargo run onitamaserver
3. Start the frontend with yarn start
4. Visit http://localhost:3000 to see the app

To develop single-player without the backend, start the frontend with

yarn start-local-ai