Skip to main content
Version: Next

Progressive Introduction Tutorial

Before we start

We will develop a simple V2EX forum client in this tutorial and you will learn the basics, concepts and some optimization techniques of Taro in this section. Learning this knowledge does not require prior knowledge of Taro, mini program development or multi-site development. Once you have this knowledge, you should be able to develop multi-applications quickly and efficiently.

Tips

This tutorial is for developers who like to learn as they go or have no prior knowledge of mini program development at all, if you prefer to learn step by step, please include the entire contents of documentation in its entirety. When you finish reading the documentation you will see that this tutorial is an asymptotic index of the documentation.

This tutorial will be divided into four chapters.

  • [Environment preparation](#Environment preparation): what needs to be installed when we use Taro.
  • [Basic tutorial](#Basic tutorial): basic concepts and development guidelines for Taro.
  • [Project Progression and Optimization](#Project Progression and Optimization): what should be done to maintain or improve the maintainability and performance of the application as the project gets bigger and slower.
  • [Multi-Ended Development](#Multi-Ended Development): How to quickly expand to a multi-ended application when you have already developed one end of the application using Taro.

(If you're browsing on a large screen) The title directory to the right of this tutorial contains subdirectories for all the chapters, so you can view or navigate to the chapter you're interested in.

Pre-requisite knowledge

In this tutorial we assume that you already have some knowledge of web front-end development and JavaScript. We will implement our application using the React and Vue frameworks respectively, switching between implementations by clicking on the React or Vue buttons in the code examples. If you are not familiar with either framework, you can cross reference the code implementation via React documentation or Vue documentation.

Our code implementation also uses some ES6 syntax, which you can see or learn about at ES6 Getting Started Tutorial

Environment Preparation

Currently Taro offers only one development method: installing the Taro command line tool (Taro CLI) for development.

The Taro CLI relies on the Node.js environment, so it must be installed on your machine. There are various ways to install the Node.js environment, if you don't know Node.js at all you can visit the Node.js official website to download an executable to install it. We recommend installing the LTS version of Node.js (currently LTS version is v12).

When your machine already has a Node.js environment, you can install the Taro CLI by entering the command npm i -g @tarojs/cli in the terminal. Once installed, enter the command taro in the terminal and if something like this appears, the installation was successful.

👽 Taro v3.0.0-beta.6

Usage: taro <command> [options]

Options:
-V, --version output the version number
-h, --help output usage information

Commands:
init [projectName] Init a project with default templete
config <cmd> Taro config
create Create page for project
build Build a project with options
update Update packages of taro
convert Convert weapp to taro
info Diagnostics Taro env info
doctor Diagnose taro project
help [cmd] display help for [cmd]

Editor

We recommend using VSCode or WebStorm (or other Jetbrains IDEs that support web development).

When you use VSCode, it is recommended to install the ESLint plugin, and if you use TypeScript, don't forget to configure eslint.probe parameter if you use TypeScript. If you use Vue, it is recommended to install the Vetur plugin.

If you are willing to spend money and are lazy you can choose WebStorm (or any other Jetbrains IDE that supports web development), which requires basically no configuration.

Whether you use VSCode or WebStrom, you can use Taro for auto-completion and live code inspection (linting) after installing the above plugin.

Terminal

macOS/Linux

What tool is used by the terminal emulator on *nix systems (Terminal/iTerm2/Konsole/Hyper/etc...) It doesn't matter, but for running the Taro CLI shell we recommend bash or zsh.

Windows

On Windows we recommend using the built-in cmd or PowerShell. If available, we recommend installing WSL and running the Taro CLI from a terminal in the Linux distribution. Windows extremes may not have been taken into account, resulting in bugs.

Seeking help

When you encounter a problem during development, you can scan the code to join WeChat Developer Group and ask a question, or go to Taro Community and ask a question. When you're sure there's a bug in Taro, feel free to ask your question to the Taro GitHub Issue.

Basic Tutorial

After installing the Taro CLI you can create a brand new project with the taro init command, you can fill in the options according to your project needs, a minimal version of a Taro project will include the following files.

├── babel.config.js             # Babel Configuration
├── .eslintrc.js # ESLint Configuration
├── config # Compile configuration directory
│ ├── dev.js # Development Mode Configuration
│ ├── index.js # Default Configuration
│ └── prod.js # Production Mode Configuration
├── package.json # Node.js manifest
├── dist # Packaged Directory
├── project.config.json # Mini Program Project Configuration
├── src # 源码目录
│ ├── app.config.js # Global Configuration
│ ├── app.css # Global CSS
│ ├── app.js # Entry Component
│ ├── index.html # H5 Entry HTML
│ └── pages # Page Component
│ └── index
│ ├── index.config.js # Page Configuration
│ ├── index.css # Page CSS
│ └── index.jsx # Page Component,If the Vue project, this file is index.vue

We will explain what each file does later, but for now, let's focus on the src folder, which is the source directory.

Entry Component

Every Taro project has an entry component and an entry configuration where we can set the global state/global lifecycle, a minimized entry component would look like this.

src/app.js
import React, { Component } from 'react'
import './app.css'

class App extends Component {
render () {
// this.props.children is the page that will be rendered
return this.props.children
}
}

// Each entry component must export a React component
export default App

Each entry component (e.g. app.js) is always accompanied by a global configuration file (e.g. app.config.js) in which we can set the path, global window, routing, etc. of the page component. A minimal global configuration is as follows

src/app.config.js
export default {
pages: [
'pages/index/index'
]
}

You may notice that the global configuration is the same for both React and Vue. This is in the configuration file, Taro doesn't care about the framework difference, the Taro CLI will execute the globally configured code directly in the Node.js environment at compile time and serialize the export default exported object to a JSON file. Next we'll talk about [page configuration](# page components) which also executes the same logic.

Therefore, we must ensure that the configuration file is executable in the Node.js environment, and not use some package or code that can only run in the H5 environment or mini program environment, otherwise the compilation will fail.

Learn more

Taro's entry component and global configuration specifications are based on WeChat mini programs and are unified for the entire platform. You can access the React entry component and Vue-entry, and global-config for details on the entry components and global configuration.

Page Component

Page components are the pages that will be rendered by each route. Taro's pages are placed in src/pages by default, and every Taro project has at least one page component. In our generated project, there is a page component: src/pages/index/index, and if you are careful, you can see that this path happens to correspond to the pages of our global configuration in the pages field. A simple page component would look like this.

src/pages/index/index.jsx
import { View } from '@tarojs/components'
class Index extends Component {
state = {
msg: 'Hello World!'
}

onReady () {
console.log('onReady')
}

render () {
return <View>{ this.state.msg }</View>
}
}

export default Index

Isn't this the familiar React and Vue components! But there are two subtle differences.

  1. onReady lifecycle function. Taro injects most of the mini program specification page lifecycle into the page component at runtime, while the lifecycle of React or Vue is also fully functional.
  2. View component. This is a cross-platform component from @tarojs/components. As opposed to the familiar div and span elements, in Taro we are going to use all such cross-platform components for development.

Like the portal component, each page component (e.g. index.vue) will also have a page configuration (e.g. index.config.js), where we can set the page's navigation bar, background color, and other parameters.

src/pages/index/index.config.js
export default {
navigationBarTitleText: 'Home'
}
Learn more

Taro's page hook functions and page configuration specifications are based on WeChat mini program and are unified for the entire platform. You can access the React page component and Vue page component for a full list of page hook functions and page configuration specifications.

Custom components

If you've made it this far, you have to be congratulated for understanding the most complex concepts in Taro: entry components and page components, and how they interact (via configuration files). The next part, if you are already familiar with React or Vue and web development, is too simple.

Let's start by writing the front page, which has a simple logic: display the latest forum posts.

src/pages/index/index.jsx
import Taro from '@tarojs/taro'
import React from 'react'
import { View } from '@tarojs/components'
import { ThreadList } from '../../components/thread_list'
import api from '../../utils/api'

import './index.css'

class Index extends React.Component {
config = {
navigationBarTitleText: 'home'
}

state = {
loading: true,
threads: []
}

async componentDidMount () {
try {
const res = await Taro.request({
url: api.getLatestTopic()
})
this.setState({
threads: res.data,
loading: false
})
} catch (error) {
Taro.showToast({
title: 'Error loading remote data'
})
}
}

render () {
const { loading, threads } = this.state
return (
<View className='index'>
<ThreadList
threads={threads}
loading={loading}
/>
</View>
)
}
}

export default Index
Learn more

You may notice that sending requests in a Taro app is done by Taro.request(). Like page configuration and global configuration, Taro's API specification is based on the WeChat mini program and is unified for the whole platform. You can find it in the API documentation to find all the APIs.

In our home page component, there is also a reference to a ThreadList component, which we will now implement.

src/components/thread_list.jsx
import React from 'react'
import { View, Text } from '@tarojs/components'
import { Thread } from './thread'
import { Loading } from './loading'

import './thread.css'

class ThreadList extends React.Component {
static defaultProps = {
threads: [],
loading: true
}

render () {
const { loading, threads } = this.props

if (loading) {
return <Loading />
}

const element = threads.map((thread, index) => {
return (
<Thread
key={thread.id}
node={thread.node}
title={thread.title}
last_modified={thread.last_modified}
replies={thread.replies}
tid={thread.id}
member={thread.member}
/>
)
})

return (
<View className='thread-list'>
{element}
</View>
)
}
}

export { ThreadList }
src/components/thread.jsx
import Taro, { eventCenter } from '@tarojs/taro'
import React from 'react'
import { View, Text, Navigator, Image } from '@tarojs/components'

import api from '../utils/api'
import { timeagoInst, Thread_DETAIL_NAVIGATE } from '../utils'

class Thread extends React.Component {

handleNavigate = () => {
const { tid, not_navi } = this.props
if (not_navi) {
return
}
eventCenter.trigger(Thread_DETAIL_NAVIGATE, this.props)
// 跳转到帖子详情
Taro.navigateTo({
url: '/pages/thread_detail/thread_detail'
})
}

render () {
const { title, member, last_modified, replies, node, not_navi } = this.props
const time = timeagoInst.format(last_modified * 1000, 'zh')
const usernameCls = `author ${not_navi ? 'bold' : ''}`

return (
<View className='thread' onClick={this.handleNavigate}>
<View className='info'>
<View>
<Image src={member.avatar_large} className='avatar' />
</View>
<View className='middle'>
<View className={usernameCls}>
{member.username}
</View>
<View className='replies'>
<Text className='mr10'>
{time}
</Text>
<Text>
评论 {replies}
</Text>
</View>
</View>
<View className='node'>
<Text className='tag'>
{node.title}
</Text>
</View>
</View>
<Text className='title'>
{title}
</Text>
</View>
)
}
}

export { Thread }

Here you can see that we have split the forum post rendering logic into two components and placed them in the src/components file, as these components are used many times in other pages. The strength of the split is entirely up to the developer, and Taro does not specify that components must be placed in the components folder, nor that pages must be placed in the pages folder.

Another point worth noting is that we don't use HTML components like div/span, but rather cross-platform components like View/Text.

Learn more

The cross-platform component library of the Taro documentation contains all component parameters and usage. However, the parameters and component names in the component library documentation are currently specific to React (except for React's click event, which is onClick). For Vue, component names and component parameters are named in a short horizontal line style (kebab-case), for example: <picker-view indicator-class="myclass" />

Router And Tabbar

In the src/components/thread component, we pass the

Taro.navigateTo({ url: '/pages/thread_detail/thread_detail' })

Jump to post details, but this page is still not implemented, now we go to the entry file to configure a new page:

src/app.config.js

export default {
pages: [
'pages/index/index',
'pages/thread_detail/thread_detail'
]
}

Then in the path src/pages/thread_detail/thread_detail to implement the post details page, the route can be jumped, and our whole process is running up to.

src/pages/thread_detail/thread_detail
import Taro from '@tarojs/taro'
import React from 'react'
import { View, RichText, Image } from '@tarojs/components'
import { Thread } from '../../components/thread'
import { Loading } from '../../components/loading'
import api from '../../utils/api'
import { timeagoInst, GlobalState } from '../../utils'

import './index.css'

function prettyHTML (str) {
const lines = ['p', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6']

lines.forEach(line => {
const regex = new RegExp(`<${line}`, 'gi')

str = str.replace(regex, `<${line} class="line"`)
})

return str.replace(/<img/gi, '<img class="img"')
}

class ThreadDetail extends React.Component {
state = {
loading: true,
replies: [],
content: '',
thread: {}
} as IState

config = {
navigationBarTitleText: 'Topic'
}

componentWillMount () {
this.setState({
thread: GlobalState.thread
})
}

async componentDidMount () {
try {
const id = GlobalState.thread.tid
const [{ data }, { data: [ { content_rendered } ] } ] = await Promise.all([
Taro.request({
url: api.getReplies({
'topic_id': id
})
}),
Taro.request({
url: api.getTopics({
id
})
})
])
this.setState({
loading: false,
replies: data,
content: prettyHTML(content_rendered)
})
} catch (error) {
Taro.showToast({
title: 'Error loading remote data'
})
}
}

render () {
const { loading, replies, thread, content } = this.state

const replieEl = replies.map((reply, index) => {
const time = timeagoInst.format(reply.last_modified * 1000, 'zh')
return (
<View className='reply' key={reply.id}>
<Image src={reply.member.avatar_large} className='avatar' />
<View className='main'>
<View className='author'>
{reply.member.username}
</View>
<View className='time'>
{time}
</View>
<RichText nodes={reply.content} className='content' />
<View className='floor'>
{index + 1} floor
</View>
</View>
</View>
)
})

const contentEl = loading
? <Loading />
: (
<View>
<View className='main-content'>
<RichText nodes={content} />
</View>
<View className='replies'>
{replieEl}
</View>
</View>
)

return (
<View className='detail'>
<Thread
node={thread.node}
title={thread.title}
last_modified={thread.last_modified}
replies={thread.replies}
tid={thread.id}
member={thread.member}
not_navi={true}
/>
{contentEl}
</View>
)
}
}

export default ThreadDetail

So far, we have implemented all the logic of this application, except for the "node list" page (we will discuss this page component in the advanced guide), and the rest of the pages can be abstracted quickly with the components or pages we have already explained. According to our plan, this application will have five pages, which are

  1. Home page, showing the latest posts (completed)
  2. Node list
  3. Top posts (can be multiplexed by component)
  4. node posts (can be multiplexed by component)
  5. post details (completed)

The first three of these pages can be planned in tabBar, which is Taro's built-in navigation bar that can be configured in app.config.js, and the page in the tabBar position will display a navigation bar after configuration. Our final app.config.js will look like this.

app.config.js
export default {
pages: [
'pages/index/index',
'pages/nodes/nodes',
'pages/hot/hot',
'pages/node_detail/node_detail',
'pages/thread_detail/thread_detail'
],
tabBar: {
list: [{
'iconPath': 'resource/latest.png',
'selectedIconPath': 'resource/lastest_on.png',
pagePath: 'pages/index/index',
text: 'Latest'
}, {
'iconPath': 'resource/hotest.png',
'selectedIconPath': 'resource/hotest_on.png',
pagePath: 'pages/hot/hot',
text: 'Hot'
}, {
'iconPath': 'resource/node.png',
'selectedIconPath': 'resource/node_on.png',
pagePath: 'pages/nodes/nodes',
text: 'Node'
}],
'color': '#000',
'selectedColor': '#56abe4',
'backgroundColor': '#fff',
'borderStyle': 'white'
},
window: {
backgroundTextStyle: 'light',
navigationBarBackgroundColor: '#fff',
navigationBarTitleText: 'V2EX',
navigationBarTextStyle: 'black'
}
}

Project Progression and Optimization

Status Management

In our implementation of the post component (src/components/thread), an event is raised via Taro's built-in eventCenter to inject the current post's data into a global GlobalState, and then the current post's data is fetched from the GlobalState on the post details page - a simple publish/subscribe pattern that is very efficient and clear when dealing with simple logic.

Once our business logic becomes complex, a simple publish-subscribe mechanism tied to a global state may cause our data flow to become difficult to track. The good thing is that there are good solutions to this problem in both the React and Vue communities. We'll use the two most popular state management tools in the community: Redux and Vuex to solve this problem.

First install redux and react-redux:

npm i redux react-redux

Inject context into our application using the Provider of react-redux in the entry file.

src/app.js
import React, { Component } from 'react'
import { Provider } from 'react-redux'
import { createStore, combineReducers } from 'redux';
import './app.css'

const reducers = combineReducers({
thread: (state = {}, action) => {
if (action.type === 'SET_CURRENT_THREAD') {
return {
...state,
...action.thread
}
}
return state
}
})

const store = createStore(reducers)

class App extends Component {
render () {
// this.props.children is the page that will be rendered
return (
<Provider store={store}>
{this.props.children}
</Provider>
)
}
}

export default App

Then in the post component we can set the current post via connect a dispatch:

src/components/thread.jsx
- eventCenter.trigger(Thread_DETAIL_NAVIGATE, this.props)
+ this.props.setThread(this.props)
- export default Thread
+ const mapDispatchToProps = dispatch => {
+ return {
+ setThread: thread => dispatch({ type: 'SET_CURRENT_THREAD', thread })
+ }
+ }
+ export default connect(null, mapDispatchToProps)(Thread)

In the post details component we get the data of the current post via connect a mapStateToProps.

src/components/thread_detail.jsx
- const id = GlobalState.thread.tid
+ const id = this.props.thread.tid
- export default ThreadDetail
+ function mapStateToProps(state) {
+ return { thread: state.thread }
+ }
+ export default connect(mapStateToProps)(ThreadDetail)
Please note

This tutorial demonstrates Redux minimalist usage, not best practice. Please visit the Redux documentation and the react-redux documentation for details.

Other status management tools

In principle, Taro can support any React- or Vue-compatible state management tool, and using such tools usually requires that the context be injected into the entry component, and in Taro the entry file cannot render the UI. Just be aware of this.

In the Vue ecosystem we recommend using Vuex. The React ecosystem has a wide variety of state management tools, and given that many developers using Taro compile their applications into mini programs, we recommend a few state management tools that have performance or size advantages.

  • mobx-react: Responsive state management tools like Vuex
  • unstaged: Minimalist state management tool based on React Hooks, with a compressed size of 200 bytes
  • Recoil: Facebook's React Hooks-based state management tool

CSS Tools

In Taro, we can freely use CSS pre-processors and post-processors, and it's as simple as adding the relevant plug-ins to the build configuration.

config/index.js
const config = {
projectName: 'v2ex',
date: '2018-8-3',
designWidth: 750,
sourceRoot: 'src',
outputRoot: 'dist',
plugins: [
'@tarojs/plugin-sass', // use Sass
// '@tarojs/plugin-less', // use Less
// '@tarojs/plugin-stylus', // use Stylus
],
defineConstants: {
},
mini: {

},
h5: {
publicPath: '/',
staticDirectory: 'static',
module: {
postcss: {
autoprefixer: {
enable: true
}
}
}
}
}

module.exports = function (merge) {
if (process.env.NODE_ENV === 'development') {
return merge({}, config, require('./dev'))
}
return merge({}, config, require('./prod'))
}
Learn more

In addition to CSS preprocessors, Taro also supports CSS Modules and CSS-in-JS. Additional CSS tools are supported in principle, and we will continue to discuss this in Custom Compilation.

Render HTML

In the post details component (ThreadDetail), we use the built-in component RichText to render HTML, but this component is not compatible and does not work properly on all ends, and some specific HTML elements cannot be rendered.

Fortunately, Taro has built-in HTML rendering, and the usage is not much different from React/Vue for web development.

src/pages/thread_detail/thread_detail.jsx
- <RichText nodes={reply.content} className='content' />
+ <View dangerouslySetInnerHTML={{ __html: reply.content }} className='content'></View>
Learn more

Taro's built-in HTML rendering features can be used not only in a web development manner, but also support advanced features such as custom styles, custom rendering, and custom events. You can visit the HTML rendering documentation to learn more.

Performance Optimization

Virtual List

In the post list component (ThreadList), we render the data directly from the remote. This is not a problem, but if we have very large data, or if the list renders an unusually complex DOM structure, this can create performance issues.

To solve this problem, Taro has a built-in VirtualList feature that renders only the view of the current visible viewport, rather than rendering all the list data in full.

src/pages/thread_detail/thread_list.jsx
import React from 'react'
import { View, Text } from '@tarojs/components'
import { Thread } from './thread'
import { Loading } from './loading'
import VirtualList from '@tarojs/components/virtual-list'


import './thread.css'

const Row = React.memo(({ thread }) => {
return (
<Thread
key={thread.id}
id={thread.id}
node={thread.node}
title={thread.title}
last_modified={thread.last_modified}
replies={thread.replies}
tid={thread.id}
member={thread.member}
/>
)
})

class ThreadList extends React.Component {
static defaultProps = {
threads: [],
loading: true
}

render () {
const { loading, threads } = this.props

if (loading) {
return <Loading />
}

const element = (
<VirtualList
height={800} /* list height */
width='100%' /* list width */
itemData={threads} /* list render data */
itemCount={threads.length} /* Length of the rendering list */
itemSize={100} /* The height of a single item in the list */
>
{Row}
</VirtualList>
)

return (
<View className='thread-list'>
{element}
</View>
)
}
}

export { ThreadList }
Learn more

In the documentation virtual-list you can find some advanced uses of virtual lists, such as: infinite scrolling, scroll offsets, scroll events, etc.

Prerender

Now let's implement the last page: the node list page. This page essentially renders a huge list that exists locally:

src/pages/nodes/nodes.jsx
import React from 'react'
import { View, Text, Navigator } from '@tarojs/components'
import allNodes from './all_node'
import api from '../../utils/api'

import './nodes.css'

function Nodes () {
const element = allNodes.map(item => {
return (
<View key={item.title} className='container'>
<View className='title'>
<Text style='margin-left: 5px'>{item.title}</Text>
</View>
<View className='nodes'>
{item.nodes.map(node => {
return (
<Navigator
className='tag'
url={`/pages/node_detail/node_detail${api.queryString(node)}`}
key={node.full_name}
>
<Text>{node.full_name}</Text>
</Navigator>
)
})}
</View>
</View>
)
})
return <View className='node-container'>{element}</View>
}

export default Nodes

This time our whole application is finished. However, if you put this app in a real mini program, especially in some real machines with low performance, it may take longer to switch to this page and there will be a white screen time.

This is due to Taro's rendering mechanism: during page initialization, native mini program can fetch data directly from local rendering, but Taro renders the initial data into a DOM tree via React/Vue, then serializes the DOM tree and gives it to the mini program to render. This means that Taro spends one more time calling the setData function on page initialization than native mini programs do - and most mini program performance problems are caused by oversized setData data.

To solve this problem, Taro has introduced a technology called Prerender, which, like server-side rendering, converts the pages to be rendered directly into wxml strings in the Taro CLI, resulting in the same or even faster speeds than native mini program.

Using prerender is also very simple, we just need to do a simple configuration:

config/prod.js
const config = {
...
mini: {
prerender: {
include: ['pages/nodes/nodes'], // `pages/nodes/nodes`
}
}
};

// We only turn on pre-rendering here when we compile for production mode
// If you need to turn it on during development, then put the configuration in `config/index` or `config/dev`
module.exports = config
Learn more

The configuration of prerender supports conditional rendering pages, conditional rendering logic, custom rendering functions, etc. For details, visit the pre-rendering documentation.

Packing volume

By default using production mode packaging, Taro will optimize the packaging volume for you. However, it is worth noting that Taro's default packaging configuration is designed to work for most projects and requirements, and is not optimal for any project. So you can build on top of the Taro configuration and optimize it for your own project.

JavaScript

In Taro applications, all Java(Type)Script is configured via babel.config.js, specifically using the babel-prest-taro Babel plugin is compiled.

By default Taro is compatible with all syntaxes supported by @babel/preset-env and is compatible with iOS 9 and Android 5, but if you don't need that high compatibility or don't need If you don't need that much compatibility or don't need some ES2015+ syntax support, you can configure babel.config.js yourself to reduce the package size.

For example, we can upgrade the compatibility to iOS 12.

babel.config.js
// babel.config.js
module.exports = {
presets: [
['taro', {
targets: {
ios: '12'
}
}]
]
}

You can visit the Babel documentation for more information on customizing your configuration.

Packaged volume analysis

Taro uses Webpack as its internal packaging system, and sometimes Webpack doesn't give us the tree-shaking effect when our business code uses the require syntax or the import default syntax. . In such cases we use webpack-bundle-analyzer to analyze our dependency packaging volume, and this plugin opens a visual graphical page in our browser telling us reference the volume of each package.

First install the webpack-bundle-analyzer dependency:

npm install webpack-bundle-analyzer -D

And then addd the following configuration to the mini.webpackChain

config/index
const config = {
...
mini: {
webpackChain (chain, webpack) {
chain.plugin('analyzer')
.use(require('webpack-bundle-analyzer').BundleAnalyzerPlugin, [])
}
}
}

After running the compile command you can see the file dependencies and volumes.

You can visit the webpack-bundle-analyzer documentation for detailed usage.

SubPackages

In some cases, we want our pages to be loaded on demand only when they are used. This is called subpackaging in the Taro application, and it is very simple to use, just by configuring the entry file app.config.js.

Suppose we need to subpackage all the node pages that we have just implemented for pre-rendering.

src/app.config.js
export default {
pages: [
'pages/index/index',
// 'pages/nodes/nodes', Remove the pages to be subpackaged from the `pages` field
'pages/hot/hot',
'pages/node_detail/node_detail',
'pages/thread_detail/thread_detail'
],
// Add subpackages to the `subpackages` field, and if the target is an Alipay mini program, you need to add a field `subPackages` with the same value as `subpackages`
// Can't be in the `pages` root directory nor outside the pages directory, you need to create a new folder in the `pages` root directory for subpackages
"subpackages": [
{
"root": "pages",
"pages": [
"nodes/nodes"
]
}
]
tabBar: {
list: [{
'iconPath': 'resource/latest.png',
'selectedIconPath': 'resource/lastest_on.png',
pagePath: 'pages/index/index',
text: 'latest'
}, {
'iconPath': 'resource/hotest.png',
'selectedIconPath': 'resource/hotest_on.png',
pagePath: 'pages/hot/hot',
text: 'hot'
}, {
// 如果是分包的子页面,就不能在 `tabBar` 中使用
'iconPath': 'resource/node.png',
'selectedIconPath': 'resource/node_on.png',
pagePath: 'pages/nodes/nodes',
text: 'node'
}],
'color': '#000',
'selectedColor': '#56abe4',
'backgroundColor': '#fff',
'borderStyle': 'white'
},
window: {
backgroundTextStyle: 'light',
navigationBarBackgroundColor: '#fff',
navigationBarTitleText: 'V2EX',
navigationBarTextStyle: 'black'
}
}

Custom Compilation

In specific cases where Taro's own compilation system does not meet our compilation needs, Taro offers two options for extending the compilation.

Extending with Webpack

In [package volume analysis](#package volume analysis) we added a Webpack plugin to mini.webpackChain and added a Webpack plugin to mini.webpackChain to achieve the effect of packing volume/dependency analysis.

In fact with the mini.webpackChain configuration we can use almost any plugin and loader from the Webpack ecosystem, for example if we want to use CoffeeScript for development.

config/index
const config = {
...
mini: {
webpackChain (chain, webpack) {
chain.merge({
module: {
rule: {
test: /\.coffee$/,
use: [ 'coffee-loader' ]
}
}
})
}
}
}

Similarly, the CSS Modules that we mentioned before can be extended with support in the form of Webpack. For more information, visit the webpack-chain documentation for details on how to use it.

Expansion with plugin system

At [CSS Tools](#CSS Tools) we have used a plugin called @tarojs/plugin-sass to implement support for Sass. Instead of using Webpack extensions to compile, Taro's plugin functionality eliminates the need to configure Webpack on each end and just uses the plugin.

In addition, Taro's plugin functionality extends the Taro CLI compilation commands, the compilation process, and the compilation platform, and you can visit the plug-in functionality documentation for more information on custom configuration.

to learn more

In addition to the above two methods, Taro also offers a number of compile-related options, which you can learn more about by visiting the compile configuration details documentation to learn more.

Multi-Ended Development

Cross-platform development

In some cases, the performance or business logic of different platforms is qualitatively different. In such cases, there is no way for us to do "one set of code for all".

For example, if we are implementing a V2EX forum application, the current API has no way to be called across domains in the browser, so we need to use another copy of the API on the H5 side, which we can solve by using the built-in environment variables.

- import api from '../../utils/api'
// We can introduce different APIs depending on the platform
+ let api
+ if (process.env.TARO_ENV === 'weapp') {
+ api = require('../../utils/api-weapp')
+ } else if (process.env.TARO_ENV === 'h5') {
+ api = require('../../utils/api-h5')
+ }

Taro also provides unified interface for multi-terminal files,Finding dependencies by different nomenclature, in such cases we can keep.

import api from '../../utils/api'

statement as is, modify our file structure by adding the name of the platform between the file name and the suffix name:

.
└── utils
├── api.h5.js
├── api.weapp.js
└── index.js
Learn more

In addition to the "built-in environment variables" and "unified interface for multiple files", Taro also offers other cross-platform development solutions, which you can learn more about by visiting the documentation cross-platform development to learn more.

Synchronized debugging

By default, Taro puts all the packaged files on each side in the dist directory. If you want to synchronize debugging on multiple ends, the files compiled first will be overwritten by the files compiled later.

However, we can synchronize debugging by modifying the outputRoot of the compilation configuration.

config/index.js
const config = {
outputRoot: `dist/${process.env.TARO_ENV}`
}

Under this configuration, the compiled directory of Wechat mini programs will be dist/weapp and the compiled directory of H5 will be dist/h5.

Using native mini program components

In some cases we need to reuse the existing ecology of the mini program, and the components/libraries of the mini program are usually written for the specific mini program and can't be used directly on Taro, requiring some additional operations.

For example, in our forum application, the post details may be returned in MarkDown format on the server side, so we need towxml to render our posts.

First we need to reference towxml in the configuration file of the post details page:

export default {
"usingComponents": {
"towxml":"../../towxml/towxml"
}
}

Then using the towxml component, it's important to remember that whether it's React or Vue, the native mini program component declaration needs to be lowercase

src/pages/thread_detail/thread_detail.jsx
- <View dangerouslySetInnerHTML={{ __html: reply.content }} className='content'></View>
+ <towxml nodes="{{reply.content}}" />

Finally, just follow the towxml document to call it.

Please note

Once the native mini program component is used, the Taro app loses its cross-over capabilities.

Learn more

Taro also supports the use of mini program plugins, for details visit the documentation Using native third-party components and plugins for mini programs.