Skip to content

Latest commit

 

History

History
237 lines (192 loc) · 18.4 KB

FAQ.md

File metadata and controls

237 lines (192 loc) · 18.4 KB

##FAQ ###Why does this exist? This starter kit implements best practices like testing, minification, bundling, and so on. It codifies a long list of decisions that you no longer have to make to get rolling. It saves you from the long, painful process of wiring it all together into an automated dev environment and build process. It's also useful as inspiration for ideas you might want to integrate into your current development environment or build process.

###What do the scripts in package.json do? Unfortunately, scripts in package.json can't be commented inline because the JSON spec doesn't support comments, so I'm providing info on what each script in package.json does here.

Script Description
remove-demo Removes the demo application so you can begin development.
prestart Runs automatically before start. Calls remove-dist script which deletes the dist folder. This helps remind you to run the build script before committing since the dist folder will be deleted if you don't. ;)
start Runs tests, lints, starts dev webserver, and opens the app in your default browser.
lint:tools Runs ESLint on build related JS files. (eslint-loader lints src files via webpack when npm start is run)
clean-dist Removes everything from the dist folder.
remove-dist Deletes the dist folder.
create-dist Creates the dist folder and the necessary subfolders.
prebuild Runs automatically before build script (due to naming convention). Cleans dist folder, builds html, and builds sass.
build Bundles all JavaScript using webpack and writes it to /dist.
test Runs tests (files ending in .spec.js) using Mocha and outputs results to the command line. Watches all files so tests are re-run upon save.
test:cover Runs tests as described above. Generates a HTML coverage report to ./coverage/index.html
test:cover:travis Runs coverage as described above, however sends machine readable lcov data to Coveralls. This should only be used from the travis build!

Can you explain the folder structure?

.
├── .babelrc                  # Configures Babel
├── .editorconfig             # Configures editor rules
├── .eslintrc                 # Configures ESLint
├── .gitignore                # Tells git which files to ignore
├── .istanbul.yml             # Configure istanbul code coverage
├── .npmrc                    # Configures npm to save exact by default
├── README.md                 # This file.
├── dist                      # Folder where the build script places the built app. Use this in prod.
├── package.json              # Package configuration. The list of 3rd party libraries and utilities
├── src                       # Source code
│   ├── actions               # Flux/Redux actions. List of distinct actions that can occur in the app.  
│   ├── components            # React components
│   ├── constants             # Application constants including constants for Redux
│   ├── containers            # Top-level React components that interact with Redux
│   ├── favicon.ico           # favicon to keep your browser from throwing a 404 during dev. Not actually used in prod build.
│   ├── index.ejs             # Template for homepage
│   ├── index.js              # Entry point for your app
│   ├── reducers              # Redux reducers. Your state is altered here based on actions
│   ├── store                 # Redux store configuration
│   ├── styles                # CSS Styles, typically written in Sass
│   └── utils                 # Plain old JS objects (POJOs). Pure logic. No framework specific code here.
├── tools                     # Node scripts that run build related tools
│   ├── setup                 # **NEEDS DOCUMENTATION**
│   │   ├── setup.js          # **NEEDS DOCUMENTATION**
│   │   ├── setupMessage.js   # **NEEDS DOCUMENTATION**
│   │   └── setupPrompts.js   # **NEEDS DOCUMENTATION**
│   ├── build.js              # Runs the production build
│   ├── chalkConfig.js        # Centralized configuration for chalk (adds color to console statements)
│   ├── distServer.js         # Starts webserver and opens final built app that's in dist in your default browser
│   ├── nodeVersionCheck.js   # **NEEDS DOCUMENTATION**
│   ├── removeDemo.js         # **NEEDS DOCUMENTATION**
│   ├── srcServer.js          # Starts dev webserver with hot reloading and opens your app in your default browser
│   ├── startMessage.js       # **NEEDS DOCUMENTATION**
│   └── testSetup.js          # Configures mocha
├── webpack.config.dev.js     # Configures webpack for development builds
└── webpack.config.prod.js    # Configures webpack for production builds

What are the dependencies in package.json used for?

Dependency Use
autoprefixer Automatically adds vendor prefixes, using data from Can I Use.
connect-history-api-fallback Support reloading deep links
object-assign Polyfill for Object.assign
react React library
react-dom React library for DOM rendering
react-redux Redux library for connecting React components to Redux
react-router React library for routing
redux Library for unidirectional data flows
redux-thunk Middleware for redux that allows actions to be declared as functions
babel-cli Babel Command line interface
babel-core Babel Core for transpiling the new JavaScript to old
babel-loader Adds Babel support to Webpack
babel-plugin-react-display-name Add displayName to React.createClass calls
babel-plugin-transform-react-constant-elements Performance optimization: Hoists the creation of elements that are fully static to the top level. reduces calls to React.createElement and the resulting memory allocations. More info
babel-preset-latest Babel preset for ES2015, ES2016 and ES2017
babel-preset-react-hmre Hot reloading preset for Babel
babel-preset-react Add JSX support to Babel
babel-preset-stage-1 Include stage 1 feature support in Babel
browser-sync Supports synchronized testing on multiple devices and serves local app on public URL
chai Assertion library for use with Mocha
chalk Adds color support to terminal
cross-env Cross-environment friendly way to handle environment variables
css-loader Add CSS support to Webpack
enzyme Simplified JavaScript Testing utilities for React
eslint Lints JavaScript
eslint-loader Adds ESLint support to Webpack
eslint-plugin-react Adds additional React-related rules to ESLint
eslint-watch Wraps ESLint to provide file watch support and enhanced command line output
extract-text-webpack-plugin Extracts CSS into separate file for production build
file-loader Adds file loading support to Webpack
html-webpack-plugin Generates custom index.html for each environment as part of webpack build
mocha JavaScript testing library
node-sass Adds SASS support to Webpack
npm-run-all Run multiple scripts at the same time
postcss-loader Adds PostCSS support to Webpack
react-addons-test-utils Adds React TestUtils
rimraf Delete files
sass-loader Adds Sass support to Webpack
sinon Standalone test spies, stubs and mocks for JavaScript
sinon-chai Extends Chai with assertions for the Sinon.JS mocking framework
style-loader Add Style support to Webpack
webpack Bundler with plugin system and integrated development server
webpack-dev-middleware Used to integrate Webpack with Browser-sync
webpack-hot-middleware Use to integrate Webpack's hot reloading support with Browser-sync
webpack-md5-hash Hash bundles, and use the hash for the filename so that the filename only changes when contents change
yargs Easily parse command-line arguments

Where are the files being served from when I run npm start?

Webpack serves your app in memory when you run npm start. No physical files are written. However, the web root is /src, so you can reference files under /src in index.html. When the app is built using npm run build, physical files are written to /dist and the app is served from /dist.

Where is index.html?

It's generated by webpack using htmlWebpackPlugin. This plugin dynamically generates index.html based on the configuration in webpack.config. It also adds references to the JS and CSS bundles using hash-based filenames to bust cache. Separate bundles for vendor and application code are created and referencing within the generated index.html file so that vendor libraries and app code can be cached separately by the browser. The bundle filenames are based on the file's hash, so the filenames only change when the file contents change. For more information on this, read Long-term caching of static assets with Webpack and html-webpack-plugin

How is Sass being converted into CSS and landing in the browser?

Magic! Okay, more specifically, we're handling it differently in dev (npm start) vs prod (npm run build)

When you run npm start:

  1. The sass-loader compiles Sass into CSS
  2. Webpack bundles the compiled CSS into bundle.js. Sounds odd, but it works!
  3. bundle.js contains code that loads styles into the <head> of index.html via JavaScript. This is why you don't see a stylesheet reference in index.html. In fact, if you disable JavaScript in your browser, you'll see the styles don't load either.

The approach above supports hot reloading, which is great for development. However, it also create a flash of unstyled content on load because you have to wait for the JavaScript to parse and load styles before they're applied. So for the production build, we use a different approach:

When you run npm run build:

  1. The sass-loader compiles Sass into CSS
  2. The extract-text-webpack-plugin extracts the compiled Sass into styles.css
  3. buildHtml.js adds a reference to the stylesheet to the head of index.html.

For both of the above methods, a separate sourcemap is generated for debugging Sass in compatible browsers.

I don't like the magic you just described above. I simply want to use a CSS file.

No problem. Reference your CSS file in index.html, and add a step to the build process to copy your CSS file over to the same relative location /dist as part of the build step. But be forwarned, you lose style hot reloading with this approach.

I just want an empty starter kit.

This starter kit includes an example app so you can see how everything hangs together on a real app. When you're done reviewing it, run this to remove the demo app:

npm run remove-demo

Don't want to use Redux? See the next question for some steps on removing Redux.

Do I have to use Redux?

Nope. Redux is useful for applications with more complex data flows. If your app is simple, Redux is overkill. Remove Redux like this:

  1. Run npm run remove-demo
  2. Uninstall Redux related packages: npm uninstall redux react-redux redux-thunk
  3. Create a new empty component in /components.
  4. Call render on the new top level component you created in step 3 in src/index.js.

How do I remove React Router?

  1. Uninstall React Router and routing related packages: npm uninstall --save react-router
  2. Delete the following files: src/routes.js
  3. Remove import { Link, IndexLink } from 'react-router'; from top of src/components/App.js, add a reference to src/components/FuelSavingsForm.js, and replace body of (implicit) render with this: <FuelSavingsPage />.

How do I deploy this?

npm run build. This will build the project for production. It does the following:

  • Minifies all JS
  • Sets NODE_ENV to prod so that React is built in production mode
  • Places the resulting built project files into /dist. (This is the folder you'll expose to the world).

Why are test files placed alongside the file under test (instead of centralized)?

Streamlined automated testing is a core feature of this starter kit. All tests are placed in files that end in .spec.js. Spec files are placed in the same directory as the file under test. Why?

  • The existence of tests is highly visible. If a corresponding .spec file hasn't been created, it's obvious.
  • Easy to open since they're in the same folder as the file being tested.
  • Easy to create new test files when creating new source files.
  • Short import paths are easy to type and less brittle.
  • As files are moved, it's easy to move tests alongside.

That said, you can of course place your tests under /test instead, which is the Mocha default. If you do, you can simplify the test script to no longer specify the path. Then Mocha will simply look in /test to find your spec files.

How do I debug?

Since browsers don't currently support ES6, we're using Babel to compile our ES6 down to ES5. This means the code that runs in the browser looks different than what we wrote. But good news, a sourcemap is generated to enable easy debugging. This means your original JS source will be displayed in your browser's dev console. Note: When you run npm start, no JS is minified. Why? Because minifying slows the build. So JS is only minified when you run the npm run build script. See more on building for production below.

Also note that no actual physical files are written to the filesystem during the dev build. For performance, all files exist in memory when served from the webpack server.. Physical files are only written when you run npm run build.

Tips for debugging via sourcemaps:

  1. Browsers vary in the way they allow you to view the original source. Chrome automatically shows the original source if a sourcemap is available. Safari, in contrast, will display the minified source and you'll have to cmd+click on a given line to be taken to the original source.
  2. Do not enable serving files from your filesystem in Chrome dev tools. If you do, Chrome (and perhaps other browsers) may not show you the latest version of your code after you make a source code change. Instead you must close the source view tab you were using and reopen it to see the updated source code. It appears Chrome clings to the old sourcemap until you close and reopen the source view tab. To clarify, you don't have to close the actual tab that is displaying the app, just the tab in the console that's displaying the source file that you just changed.
  3. If the latest source isn't displaying the console, force a refresh. Sometimes Chrome seems to hold onto a previous version of the sourcemap which will cause you to see stale code.

Why does the build use npm scripts instead of Gulp or Grunt?

In short, Gulp is an unnecessary abstraction that creates more problems than it solves. Here's why.

Why does package.json reference the exact version?

This assures that the build won't break when some new version is released. Unfortunately, many package authors don't properly honor Semantic Versioning, so instead, as new versions are released, I'll test them and then introduce them into the starter kit. But yes, this means when you do npm update no new dependencies will be pulled down. You'll have to update package.json with the new version manually.

How do I handle images?

Via Webpack's file loader. Example:

<img src={require('./src/images/myImage.jpg')} />

Webpack will then intelligently handle your image for you. For the production build, it will copy the physical file to /dist, give it a unique filename, and insert the appropriate path in your image tag.

I'm getting an error when running npm install: Failed to locate "CL.exe"

On Windows, you need to install extra dependencies for browser-sync to build and install successfully. Follow the getting started steps above to assure you have the necessary dependencies on your machine.

I can't access the external URL for Browsersync

To hit the external URL, all devices must be on the same LAN. So this may mean your dev machine needs to be on the same Wifi as the mobile devices you're testing.

What about the Redux Devtools?

Install the Redux devtools extension in Chrome Developer Tools. If you're interested in running Redux dev tools cross-browser, Barry Staes created a branch with the devtools incorporated.

Hot reloading isn't working!

Hot reloading doesn't always play nicely with stateless functional components at this time. This is a known limitation that is currently being worked. To avoid issues with hot reloading for now, use a traditional class-based React component at the top of your component hierarchy.

How do I setup code coverage reporting?

Using the npm run test:cover command to run the tests, building a code coverage report. The report is written to coverage/index.html. Slingshot provides a script for this:

npm run open:cover

You can add code coverage metrics to your README.md file and pull by integrating with Coveralls.

  1. Sign in to Coveralls with your GitHub account.
  2. Authorise Coveralls to access your repositories.
  3. Choose 'Add Repo' and select your repo.

That's it! Travis will now execute the npm run test:cover:travis script after a successful build, which will write the coverage report in the standard lcov format and send it directly to Coveralls. The environment variables provided for travis jobs are used to automatically target the correct Coveralls project, as long as it is set up as described above.

You can get the badge from the Coveralls website.

###What about TypeScript? Here's a fork with TS support: