Skip to content
No description, website, or topics provided.
Haskell Scala JavaScript Shell Java CSS HTML
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Board Game Application


Version 0.9.5 beta.

The Board Game

The board game is in the same genre as other well-known crossword games: drag and drop letters to form words on a square board.

See the project web site.


This project is intended as a test bed for the development and deployment of production quality applications by using different programming languages, their ecosystems, and devops tools.

It defines a board game API, and client and server implementations for it. At this time, the API has two server implementations, one in Haskell and one in Scala, and one client implementation, in Javascript using the React framework.

For a simple representation of the API, see the Haskell module BoardGame.Common.GameApi in the haskell-server hierarchy.

The first implementation is being developed on the MAC OS/X 10.9+, deployed on a Linux Amazon EC2 instance, and accessed through modern browsers.


  • haskell-server - The game server in Haskell.

  • scala-server - The game server in Scala.

  • web-ui - The Javascript/React user interface for the game.

  • java-client - A client-side library for accessing the API in Java clients.

  • benchmark - A benchmark (in Java) simulating concurrent game sessions.

Getting Started in Development

The steps needed to set up the development and deployment environments for the application are scripted in the Dockerfiles used for packaging and deployment of the application in the docker directories of haskell-server and scala-server.

To get started with development, follow the same steps listed in the Dockerfiles on your development machine. You may start by consulting the file in the docker directory, and then reviewing the Dockerfiles there.

For Java development (java-client and benchmark sub-projects) you'll also need to install maven.

For postgres installation on ubuntu see:

After cloning the repository:

  • cp - this is the environment-setting script for development
  • edit as appropriate for your minimal environment
  • set your environment: .

Utility libraries used by the Java sub-projects are hosted on jCenter ( Make sure your maven settings.xml in $HOME/.m2 configures that repository in its profiles. The source for these libraries is open at my github (azadbolour).


Please refer to the README file in the dict/ directory. One issue to be aware of is that diectionaries are preprocessed to a list of masked words, words with blank holes in them. Because masked word pre-processing is time-consuming, the masked words file is saved in git, but in zipped form because it is too large for github.

At this time, unzipping this file for the first time, and maintaining it in case the dictionary is updated remains a manual step. Use the script dict/

Working with the Client API Library

The java-client sub-project uses a standard Maven pom project structure. See in the java-client directory.

See sample test class: com.bolour.boardgame.integration.BasicGameClientTest.

The java client code is based on Java 8.


The benchmark sub-project is written in Java, and uses a standard Maven pom structure. See and in the benchmark directory. To configure the benchmark, copy benchmark-config.yml and edit as appropriate. Then start the benchmark giving it your copy of the config file as a parameter.

The benchmark is based Java 8.


Following are some benchmark results for a version of the game circa late 2017. Latest performance benchmarks TBD.

machine: MAC-Book Pro i7 2.8GHz 4-core
dictionary: 114K words
users: 75
average think time: 100 milliseconds

haskell: average latency of a machine move: 270 milliseconds
scala: average latency of a machine move: 59 milliseconds

Since human interactions have much longer think times, scaling to hundreds of concurrent games on a single machine should not be an issue. Beyond that the application is easily amenable to horizontal scaling. The bottleneck for very high scalability may end up being the database. But at this stage of the project database interactions are de-emphasized, and very high scalability is beyond our scope.

Github Site

General Conventions

  • For trivial code cleanup and unimportant changes with minimal impact to users, TODO comments are added at the right places in the source code. The backlog lists larger tasks or tasks that cannot yet be localized in the code base.

Branches of Development

  • master - stable version
  • dev - development version - not necessarily stable - off master
  • others - temporary feature branches


Thanks to Eitan Chatav for advice on Haskell development.

Thanks to Allen Haim for advice on Javascript development.

Thanks to Dennis Allard for hosting the application at

Thanks to the Moby project for the English dictionary:




You can’t perform that action at this time.