Mancala Server

Server used to match-make different mancala bots via HTTP.

Requirements

• GNU Make: This is optional, and is used to easily build and run the program. This is not strictly speaking required, and methods of building and running are described in the "Running" section.
• sqlite: This is required, as all persistant data will be stored in a sqlite database.
• rust: The rust compiler, it's build tool cargo and several other bundled binaries are required to run the server. This project is built in rust, and this repo only contains the source code, so it can not be used without this dependency.
• trunk: This is optional, and must be present for the frontend to be built. It also requires rust and cargo to be installed on your machine.

Installation

Run git clone https://github.com/erg352/mancala-server.git --recursive-submodules. If the recursive-submodules path was not specified during the git clone, simply run git submodule init and git submodule update.

Running

Make sure the repo is properly installed, including it's submodules. Run make build-release to build the project, an make run-release to build and run it.

If GNU Make is not installed on the machine, the following commands can also be ran:

Building:

cargo build --release
# Required if we want to build the frontend as well, but
# can be omitted if this is not needed. This requires trunk,
# see the dependencies section for more information.
(cd frontend && trunk build --release)

Running:

# If we are intent on using the frontend generated by trunk
cargo run --release -- --port $(MY_PORT) --database $(MY_DATABASE_PATH) --static-routes frontend/dist

# If we do not care about the frontend.
cargo run --release -- --port $(MY_PORT) --database $(MY_DATABASE_PATH)

Where MY_PORT and MY_DATABASE_PATH corresponding to the desired port and sqlite database file the server should bind to. Run cargo run release -- --help for more information about the specific arguments the program can take.

Instead of running cargo run --release --, which will automaticaly rebuild the project (if any changes are found) each time, we can instead simply call the executable found in target/release/match-server. This can be symlinked to anywhere that's more practical to access. If we run the program this way, the -- found after --release must be omitted, as it's simply used by cargo to split cargo's commands with the called program's commands.