Use a common pattern for filename content hashes

Adding the content hash suffix to file names helps browsers cache the assets and only request the asset from the server when the content changes. This method improves the experience for returning users who only need to download the changed assets.

When comparing asset names between different builds, bundle-stats extracts the content hash and compares the assets by source file name. The most common supported patterns are:

copy
file-HASH.js
file.HASH.js

To configure asset file names, you can set the following options for the production mode config:

copy
module.exports = {
  output: {
    // Configure entry file names
    filename: '[name].[contenthash].js',
    // Configure chunk file names
    chunkFilename: '[name].[chunkhash].js',
    // Configure asset file names
    assetModuleFilename: '[path][name].[contenthash][ext][query]'
  },
  plugins: [
    // CSS file name pattern when using mini-css-extract-plugin
    new MiniCssExtractPlugin({
      filename: '[name].[contenthash].css'
    })
  ]
}

copy
assets by path *.js 10.5 KiB
  asset vendors.f1eef34c98603711f45e.js 8.71 KiB [emitted] [immutable] [minimized] (name: vendors) (id hint: vendors)
  asset main.7c920d907e4c753d0935.js 1.79 KiB [emitted] [immutable] [minimized] (name: main)
asset images/logo.31b9235056a3dc35c4c5.png 6.4 KiB [emitted] [immutable] [from: images/logo.png] (auxiliary name: main)
asset index.html 364 bytes [emitted]
asset main.b231cc1c6e07b9d4ff06.css 122 bytes [emitted] [immutable] (name: main)

copy
module.exports = {
  output: {
    // Configure entry file names
    filename: '[name].[contenthash].js',
    // Configure chunk file names
    chunkFilename: '[name].[chunkhash].js'
  },
  module: {
    rules: [
      // file name pattern when using file-loader
      {
        test: /\.(png|jpe?g|gif)$/i,
        loader: 'file-loader',
        options: {
          name: '[path][name].[contenthash].[ext]'
        }
      }
    ]
  },
  plugins: [
    // CSS file name pattern when using mini-css-extract-plugin
    new MiniCssExtractPlugin({
      filename: '[name].[contenthash].css'
    })
  ]
}

copy
Asset       Size  Chunks                         Chunk Names
images/logo.31b9235056a3dc35c4c5d9806ab46e4f.png    6.4 KiB          [emitted] [immutable]
                                      index.html  336 bytes          [emitted]
                   main.63b96e941d400fc1b806.css  122 bytes       0  [emitted] [immutable]  main
                    main.d7df904f48c9631839e8.js   1.75 KiB       0  [emitted] [immutable]  main
                 vendors.0e025fde233b62c30823.js   8.72 KiB       1  [emitted] [immutable]  vendors

You can adjust the hash length globally using output.hashDigestLength option (default 20). bundle-stats hash replacement supports a minimum of 5 characters.

Name asynchronous chunks

Asynchronous chunks defer code loading/execution, decrease the initial JS/CSS size and improve the loading performance. By default, webpack uses the chunk ID for the chunk file name. When the chunk ID changes, the chunk name will be different, forcing the users to re-download the asset even if there are no code changes.

To output predictable asset names for chunks, use webpack annotation for async chunks:

copy
const HomePageComponent = React.lazy(
  () => import(/* webpackChunkName: 'home' */'./pages/home')
);

copy
const DetailsPage = Loadable({
  loader: () => import(/* webpackChunkName: 'details' */'./pages/details'),
  loading: Loading
});

Use deterministic chunkIds (webpack 4)

By default, webpack 4 generates chunk IDs based on their order of appearance('natural' algorithm). The chunk IDs map can change across related chunks when the order changes or a chunk is added/removed. The change is causing unnecessary cache invalidation and does not allow RelativeCI to compare the chunk modules between different builds.

To generate deterministic chunk ids between different builds, you can use optimization.chunkIds option:

copy
module.exports = {
  optimization: {
    chunkIds: 'named'
  }
}