Have you ever been sitting in lecture, bored out of your mind? Well sit no further! Now introducing RollRace, an online, multiplayer racing game that soothes your itch for a good time.
This repository combines the client and server into a single repository that can be co-developed, tested and ultimately deployed to Heroku or basin.cs.middlebury.edu.
The client was created with create-react-app (CRA) and the server is a separate Node.js application. The client-server integration is based on this tutorial and repository. This repository will be referred to as the "top-level" to distinguish it from the client and server.
The skeleton is structured as three separate packages. That is a "top-level" package and a separate "client" and "server". Thus initially installing dependencies is a 3 step process that runs "install" for each of the packages.
npm install
npm install --prefix client
npm install --prefix server
The --prefix option treats the supplied path as the package root. In this case it is equivalent to cd client then npm install then cd ...
You will typically not need to install any dependencies in the top-level package.json file. Doing so is a common mistake. Most dependencies are needed by the client or the server and should be installed in the respective sub-packages, e.g. to install reactstrap for your client application:
npm install --save reactstrap --prefix client
In addition to installing dependencies, if you wish to run the application locally, you must run "npx knex migrate:latest" to intiallize the SQLITE DB.
The combined application, client and server, can be run with npm start in the top-level directory. npm start launches the CRA development server on http://localhost:3000 and the backend server on http://localhost:3001. By setting the proxy field in the client package.json, the client development server will proxy any unrecognized requests to the server. By default this starts the server in hot-reloading mode (like with the client application).
The client application can be independently tested as described in the CRA documentation, i.e.:
npm test --prefix client
The server can be similarly independently tested:
npm test --prefix server
Both the client and server can be independently linted via:
npm run lint --prefix client
and
npm run lint --prefix server
To ensure consistent style we use the CRA-recommended Prettier package. We installed it in the "top-level" package with
npm install --save-dev husky lint-staged prettier
and added the recommended configuration to automatically reformat code during the commit. That is whenever you commit your code, Prettier will automatically reformat your code during the commit process (as a "hook"). The hook is specified in the top-level package.json file. The client and server has its own ESLint configuration.
We added custom ESLint rules to capture common errors. To ensure compatibility with Prettier, we also installed the eslint-config-prettier package in both the client and server to disable styling rules that conflict with Prettier.
npm install --save-dev eslint-config-prettier --prefix server
npm install --save-dev eslint-config-prettier --prefix client
and added an "extends" entry to .eslintrc.json.
The skeleton is setup for CI with Travis-CI. Travis will build the client and test and lint both the client and the server.
Link to Heroku App: https://rollrace.herokuapp.com/
Your application can be deployed to Heroku using the approach demonstrated in this repository. The key additions to the top-level package.json file to enable Heroku deployment:
- Specify the node version in the
enginesfield - Add a
heroku-postbuildscript field that will install dependencies for the client and server and create the production build of the client application. - Specify that
node_modulesshould be cached to optimize build time
In addition a Procfile was added in the top-level package to start the server.
Assuming that you have a Heroku account, have installed the Heroku command line tools and committed any changes to the application, to deploy to Heroku:
-
Create the Heroku app, e.g.:
heroku apps:create -
Push to Heroku
git push heroku master
Depending on how you implement your server, you will likely need create "addons" for your database, etc. and migrate then seed your database before you deploy.
Your project can be deployed to basin.cs.middlebury.edu (where it is typically run within screen on an unused port). As with Heroku you will like need to create and seed your database before you deploy.
-
Build production assets for the client application (from the top-level directory):
npm run heroku-postbuild -
Start the server from the top-level directory (note you will need to pick a different, unused, port):
NODE_ENV=production PORT=5042 npm run start --prefix server