🍾🍾🍾 GIT CLONE LOCAL DEMO 🚀🚀🚀
Webpack 4 demo update in progress
Now supports Webpack 4 aggressive code splitting We have updated
webpack-flush-chunksto now support more complex code splitting!webpack-flush-chunksenables developers to leverage smarter and less wasteful chunking methods avaliable to developers inside of Webpack.
Use this package server-side to flush webpack chunks from React Universal Component or any package that flushes an array of rendered moduleIds or chunkNames. The preferred approach is chunkNames, as that's what babel-plugin-universal-import focuses on.
import { flushChunkNames } from 'react-universal-component/server'
import flushChunks from 'webpack-flush-chunks'
const app = ReactDOMServer.renderToString(<App />)
const { js, styles } = flushChunks(webpackStats, {
chunkNames: flushChunkNames()
})
res.send(`
<!doctype html>
<html>
<head>
${styles}
</head>
<body>
<div id="root">${app}</div>
${js}
</body>
</html>
`)This plugin allows for complex code splitting to be leveraged for improved caching and less code duplication! Below are two examples of well tested splitting configurations. If you experience any issues with bespoke optimization configurations, we would love to hear about it!
Before this update, developers were limited to a single chunk strategy. What the strategy did was give developers a similar chunk method to CommonsChunkPlugin that was used in Webpack 3. We did not support AggressiveSplittingPlugin
optimization: {
runtimeChunk: {
name: 'bootstrap'
},
splitChunks: {
chunks: 'initial',
cacheGroups: {
vendors: {
test: /[\\/]node_modules[\\/]/,
name: 'vendor'
}
}
}
},Now you can use many flexible code splitting methods, like the one below. Webpack encourages aggressive code splitting methods, so we jumped on the bandwagon and did the upgrades. Just like before, we use the chunkNames generated - then we can look within the Webpack 4 chunk graph and resolve any other dependencies or automatically generated chunks that consist as part of the initial chunk.
We can load the nested chunks in the correct order as required and if many chunks share a common chunk, we ensure they load in the correct order, so that vendor chunks are always available to all chunks depending on them without creating any duplicate requests or chunk calls.
optimization: {
splitChunks: {
chunks: 'async',
minSize: 30000,
minChunks: 1,
maxAsyncRequests: 5,
maxInitialRequests: 3,
automaticNameDelimiter: '~',
name: true,
cacheGroups: {
vendors: {
test: /[\\/]node_modules[\\/]/,
priority: -10
},
default: {
minChunks: 2,
priority: -20,
reuseExistingChunk: true
}
}
}
}The code has been cracked for while now for Server Side Rendering and Code-Splitting individually. Accomplishing both simultaneously has been an impossibility without jumping through major hoops or using a framework, specifically Next.js. Our tools are for "power users" that prefer the frameworkless approach.
Webpack Flush Chunks is essentially the backend to universal rendering components like React Universal Component. It works with any "universal" component/module that buffers a list of moduleIds or chunkNames evaluated.
Via a simple API it gives you the chunks (javascript, stylesheets, etc) corresponding to the modules that were synchronously rendered on the server, which without this package would be asynchronously rendered on the client. In doing so, it also allows your first client-side render on page-load to render async components synchronously!
This solves the problem of having to make additional requests to get async components plus React checksum mismatches when the client expects to render a <Loading /> component.
It offers 2 functions flushChunks and flushFiles, which you call immediately after ReactDOMServer.renderToString. They are used in server-rendering to extract the minimal amount of chunks to send to the client, thereby solving a missing piece for code-splitting: server-side rendering.
The dream of code-splitting everywhere is finally here.
Reactlandia Articles:
-
announcing-react-universal-component-2-and-babel-plugin-universal-import 🚀
-
how-to-use-webpack-magic-comments-with-react-universal-component
-
webpack-import-will-soon-fetch-js-and-css-heres-how-you-do-it-today
yarn add react-universal-component webpack-flush-chunks
yarn add --dev babel-plugin-universal-import extract-css-chunks-webpack-plugin
-
Babel Plugin Universal Import is used to make
react-universal-componentas frictionless as possible. It removes the need to provide additional options to insure synchronous rendering happens on the server and on the client on initial load. These packages aren't required, but usage as frictionless as possible. -
Extract Css Chunks Webpack Plugin is another companion package made to complete the CSS side of the code-splitting dream. Its also a standalone plugin thats great for codesplitting css, with built-in HMR
If you like to move fast, git clone the universal-demo. We are working on a Webpack 4 demo
React Universal Component, when used on the server, skips the loading phase and syncronously renders your contained component, while recording the ID of
its corresponding module. React Universal Component may be used multiple times and therefore may record multiple split points. flushChunks/flushFiles is then able
to determine the minimal set of chunks required to re-render those modules/components on the client. From there it outputs strings, arrays or React components
containing the precise javascript files (and CSS files) to embed in your HTML response.
The result is a server-rendered response whose "checksum" matches the one generated on the client, so that another client render is not needed, and more importantly so that another request to the server for an additional chunk is not needed.
For future imports performed on user navigation, the "dual-import" mechanism of babel-plugin-universal-import will request a stylesheet. To accomplish that, a hash of chunk names to stylsheets is provided so you can embed it in the page, similar to what webpack does with your js chunks in its bootstrap code.
Before we examine how to use flushChunks/flushFiles, let's take a look at the desired output. It's something like this:
<head>
<link rel='stylesheet' href='/static/0.css' />
<link rel='stylesheet' href='/static/7.css' />
<link rel='stylesheet' href='/static/main.css' />
</head>
<body>
<div id="react-root"></div>
<!-- before entry chunks -->
<script type='text/javascript' src='/static/bootstrap.js'></script>
<script type='text/javascript' src='/static/vendor.js'></script>
<!-- dynamic chunks -->
<script type='text/javascript' src='/static/0.js'></script>
<script type='text/javascript' src='/static/7.js'></script>
<!-- after entry chunks -->
<script type='text/javascript' src='/static/main.js'></script>
</body>Notice common
vendorandbootstrapchunks at the beginning and your main entry bundle (main) at the end. Notice that chunks0and7are served, but not chunks1-6or8+. That's a lot of bytes saved in initial requests!
Because of the way Webpack works where "bootstrap" code must be run before any additional chunks can be registered, it's imperative bootstrap and common chunks are generated and placed at the beginning, thereby allowing you to place dynamic chunks before your entry chunk which kickstarts app rendering.
In conjunction with your Webpack configuration (which we'll specify below), Webpack Flush Chunks solves these problems for you by consuming your Webpack compilation stats and generating strings, arrays and components you can embed in the final output rendered on the server.
.babelrc:
{
"plugins": ["universal-import"]
}
src/components/App.js:
import universal from 'react-universal-component'
const UniversalComponent = universal(props => import(`./${props.page})
export default () =>
<div>
<UniversalComponent page='Foo' />
</div>server/render.js:
import ReactDOMServer from 'react-dom/server'
import { flushChunkNames } from 'react-universal-component/server'
import flushChunks from 'webpack-flush-chunks'
const app = ReactDOMServer.renderToString(<App />)
const chunkNames = flushChunkNames()
const { js, styles } = flushChunks(stats, { chunkNames })
res.send(`
<!doctype html>
<html>
<head>
${styles}
</head>
<body>
<div id="root">${app}</div>
${js}
</body>
</html>
`)et voila!
Note: if you require a less automated approach where you're given just the stylesheets and scripts corresponding to dynamic chunks (e.g. not
main.js), seeflushFilesin the the low-level API section.
flushChunks(stats, {
chunkNames: ReactUniversalComponent.flushChunkNames(),
before: ['bootstrap', 'vendor'], // default
after: ['main'], // default
outputPath: path.resolve(__dirname, '../dist'), // required only if you want to serve raw CSS
})-
chunkNames - ***array of chunks flushed from
react-universal-component -
before - array of named entries that come BEFORE your dynamic chunks:
A typical pattern is to create avendorchunk. A better strategy is to create avendorand abootstrapchunk. The "bootstrap" chunk is a name provided to theCommonsChunkPluginwhich has no entry point specified for it. The plugin by default removes webpack bootstrap code from the namedvendorcommon chunk and puts it in thebootstrapchunk. This is a common pattern because the webpack bootstrap code has info about the chunks/modules used in your bundle and is likely to change, which means to cache yourvendorchunk you need to extract the bootstrap code into its own small chunk file. If this is new to you, don't worry. Below you will find examples for exactly how to specify your Webpack config. Lastly, you do not need to provide this option if you have abootstrapchunk, orvendorchunk or both, as those are the defaults.
Mostly related to Webpack 2 & 3. It is still very useful if you need to load a specific chunk name first webpack-flush-chunks now can rely on a better chunk graph provided by Webpack 4 - chunks are loaded in the correct order with more autonomy.
-
after - array of named entries that come AFTER your dynamic chunks:
Similar tobefore,aftercontains an array of chunks you want to come after the dynamic chunks that your universal component flushes. Typically you have just amainchunk, and if that's the case, you can ignore this option, as that's the default. -
outputPath - absolute path to the directory containing your client build: This is only needed if serving css embedded in your served response HTML, rather than links to external stylesheets. I.e. if you are using the
Cssandcssvalues in thereturn APIdescribed in the next section. It's needed to determine where in the file system to find the CSS that needs to be extract into an in-memory string. Keep in mind if you're rendering the server with Webpack, filesystem paths may not match up, so it's important to accurately pass theoutputPathto yourserverRendermethod. We recommend to do this by running your server express/koa/hapi/etc code via Babel and then by requiring your Webpack server bundle into it. See one of our boilerplates for an example.
The return of flushChunks provides many options to render server side requests, giving you maximum flexibility:
const {
// react components:
Js, // javascript chunks
Styles, // external stylesheets
Css, // raw css
// strings:
js, // javascript chunks
styles, // external stylesheets
css, // raw css
// arrays of file names:
scripts,
stylesheets,
// cssHash for use with babel-plugin-dual-import
cssHashRaw, // hash object of chunk names to css file paths
cssHash, // string: <script>window.__CSS_CHUNKS__ = ${JSON.stringify(cssHashRaw)}</script>
CssHash, // react component of above
// important paths:
publicPath,
outputPath
} = flushChunks(moduleIds, stats, options)Let's take a look at some examples:
client:
const ExtractCssChunks = require('extract-css-chunks-webpack-plugin')
entry: [
path.resolve(__dirname, '../src/index.js'),
],
module: {
rules: [
{
test: /\.js$/,
exclude: /node_modules/,
use: 'babel-loader',
},
{
test: /\.css$/,
use: [
ExtractCssChunks.loader,
{
loader: 'css-loader',
options: {
modules: true,
localIdentName: '[name]__[local]--[hash:base64:5]'
}
]
}
]
},
plugins: [
new ExtractCssChunks(),
...server:
module: {
rules: [
{
test: /\.js$/,
exclude: /node_modules/,
use: 'babel-loader',
},
{
test: /\.css$/,
exclude: /node_modules/,
use: {
loader: 'css-loader/locals', // notice you're using the `locals` file as your loader
options: {
modules: true,
localIdentName: '[name]__[local]--[hash:base64:5]'
}
}
}
]
}
plugins: [
new webpack.optimize.LimitChunkCountPlugin({
maxChunks: 1,
})
...- The
LimitChunkCountPluginwithmaxChunks: 1insures only one file is generated for your server bundle so it can be run synchronously.
If you're specifying externals to leave unbundled, you need to tell Webpack
to still bundle react-universal-component and webpack-flush-chunks so that they know they are running within Webpack. For example:
const externals = fs
.readdirSync(modeModules)
.filter(x => !/\.bin|react-universal-component|webpack-flush-chunks/.test(x))
.reduce((externals, mod) => {
externals[mod] = `commonjs ${mod}`
return externals
}, {})Since the webpack plugin API does not yet allow you to add to stats, the AutoDllPlugin cannot add chunks to stats. Therefore you have to embed its corresponding script manually.
For advanced users that want access to all files flushed (.js, .css or whatever else might be in there) and without named entry chunks you already know (such as bootstrap, vendor, and main), here you go:
import { flushChunkNames } from 'react-universal-component/server'
import { flushFiles } from 'webpack-flush-chunks'
const chunkNames = flushChunkNames()
const scripts = flushFiles(stats, { chunkNames, filter: 'js' })
const styles = flushFiles(stats, { chunkNames, filter: 'css' })i.e. this will get you all files corresponding to flushed "dynamic" chunks.
The only thing different with the API is that it has a filter option, and that it doesn't have before, after and outputPath options. The filter can be a file extension as a string, a regex, or a function: filter: file => file.endsWith('js').
Keep in mind, you will have to get right placing these between your bootstrap and main scripts. OR if you don't have a bootstrap script, you need to set it up so your main script doesn't actually call ReactDOM.render, and instead you put <script>window.render()</script> (where window.render() calls ReactDOM.render) after all your chunks in your markup so that by the time it's called all your chunks are loaded. In the latter case, you should put your dynamic chunks received from flushFiles after your main script so that the webpack bootstrap code now within your main script (as it regularly is) knows what to do with the additional scripts from dynamic chunks.
If what you want, instead of file names, is full-on compilation chunk objects (and any information it contains, which for 99% of most projects is unnecessary), create an issue and we'll add it. But until there is an actual need, we would like to keep the API simple.
🍾🍾🍾 faceyspacey/universal-demo 🚀🚀🚀
git clone https://github.com/faceyspacey/universal-demo.git
cd universal-demo
yarn
yarn startWe use commitizen, so run npm run cm to make commits. A command-line form will appear, requiring you answer a few questions to automatically produce a nicely formatted commit. Releases, semantic version numbers, tags, changelogs and publishing to NPM will automatically be handled based on these commits thanks to semantic-release. Be good.
Reviewing a package's tests are a great way to get familiar with it. It's direct insight into the capabilities of the given package (if the tests are thorough). What's even better is a screenshot of the tests neatly organized and grouped (you know the whole "a picture says a thousand words" thing).
Below is a screenshot of this module's tests running in Wallaby ("An Integrated Continuous Testing Tool for JavaScript") which everyone in the React community should be using. It's fantastic and has taken my entire workflow to the next level. It re-runs your tests on every change along with comprehensive logging, bi-directional linking to your IDE, in-line code coverage indicators, and even snapshot comparisons + updates for Jest! I requestsed that feature by the way :). It's basically a substitute for live-coding that inspires you to test along your journey.
- redux-first-router. It's made to work perfectly with Universal. Together they comprise our "frameworkless" Redux-based approach to what Next.js does (splitting, SSR, prefetching, and routing). People are lovin it by the way 😎