Differences to React

Preact itself is not intended to be a reimplementation of React. There are differences. Many of these differences are trivial, or can be completely removed by using preact/compat, which is a thin layer over Preact that attempts to achieve 100% compatibility with React.

The reason Preact does not attempt to include every single feature of React is in order to remain small and focused - otherwise it would make more sense to simply submit optimizations to the React project, which is already a very complex and well-architected codebase.



Main differences

The main difference when comparing Preact and React apps is that we don't ship our own Synthetic Event system. Preact uses the browser's native addEventListener for event handling internally. See GlobalEventHandlers for a full list of DOM event handlers.

For us it doesn't make sense as the browser's event system supports all features we need. A full custom event implementation would mean more maintenance overhead and a larger API surface area for us.

We've come across the following differences between React's synthetic event system and native browser events:

  • Events don't bubble through <Portal> components
  • The clear "x" button for <input type="search"> does not fire an input event in IE11 - use onSearch instead
  • Use onInput instead of onChange for <input> elements (only if preact/compat is not used)
  • Use onDblClick instead of onDoubleClick (only if preact/compat is not used)

The other main difference is that Preact follows the DOM specification more closely. An example of this is the ability to use class instead of className.

Version Compatibility

For both preact and preact/compat, version compatibility is measured against the current and previous major releases of React. When new features are announced by the React team, they may be added to Preact's core if it makes sense given the Project Goals. This is a fairly democratic process, constantly evolving through discussion and decisions made in the open, using issues and pull requests.

Thus, the website and documentation reflect React 16.x and 15.x when discussing compatibility or making comparisons.

Debug messages and errors

Our flexible architecture allows addons to enhance the Preact experience in any way they want. One of those addons is preact/debug which adds helpful warnings and errors and attaches the Preact Developer Tools browser extension, if installed. Those guide you when developing Preact applications and make it a lot easier to inspect what's going on. You can enable them by adding the relevant import statement:

import "preact/debug"; // <-- Add this line at the top of your main entry file

This is different from React which requires a bundler being present that strips out debugging messages at build time by checking for NODE_ENV != "production".

Features unique to Preact

Preact actually adds a few convenient features inspired by work in the (P)React community:

Native support for ES Modules

Preact was built with ES Modules in mind from the get go and therefore was one of the first frameworks to support them. You can load Preact via the import keyword directly in browsers without having it to pass through a bundler first.

Arguments in Component.render()

For convenience we pass this.props and this.state of a class component to the render(). Take a look at this component which uses one prop and one state property.

// Works in both Preact and React
class Foo extends Component {
  state = { age: 1 };

  render() {
    return <div>Name: {this.props.name}, Age: {this.state.age}</div>;
  }
}
Run in REPL

In Preact this can be also written like this:

// Only works in Preact
class Foo extends Component {
  state = { age: 1 };

  render({ name }, { age }) {
    return <div>Name: {name}, Age: {age}</div>;
  }
}
Run in REPL

Both snippets render the exact same thing. It's just a matter of stylistic preference.

Raw HTML attribute/property names

With Preact we follow more closely the DOM specification supported by all major browsers. One prominent difference is that you can use the standard class attribute instead of className.

// This:
<div class="foo" />

// ...is the same as:
<div className="foo" />
Run in REPL

Most Preact developers prefer to use class because it's shorter to write, but both are supported.

SVG inside JSX

SVG is pretty interesting when it comes to the names of its properties and attributes. Some properties (and their attributes) on SVG objects are camelCased (e.g. clipPathUnits on a clipPath element), some attributes are kebab-case (e.g. clip-path on many SVG elements), and other attributes (usually ones inherited from the DOM, e.g. oninput) are all lowercase.

Preact forwards SVG-Attributes as is. This allows you to copy and paste unmodified SVG snippets right into your code and have them work out of the box. This allows greater interoperability with tools designers tend to use to generate icons or SVG illustrations.

If you're coming from React you're likely used to specify every attribute in camelCase. If you'd like to continue using the camelCase'd attribute names you can use our preact/compat compatibility layer. It mirrors the React API and normalizes these attributes.

// React
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 48 48">
  <circle fill="none" strokeWidth="2" strokeLinejoin="round" cx="24" cy="24" r="20" />
</svg>
// Preact (note stroke-width and stroke-linejoin)
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 48 48">
  <circle fill="none" stroke-width="2" stroke-linejoin="round" cx="24" cy="24" r="20" />
</svg>
Run in REPL

Use onInput instead of onChange

Largely for historical reasons, the semantics of React's onChange event are actually the same as the onInput event provided by browsers, which is supported everywhere. The input event is the best-suited event for the majority of cases where you want to react when a form control is modified. In Preact core, onChange is the standard DOM change event that gets fired when an element's value is committed by the user.

// React
<input onChange={e => console.log(e.target.value)} />

// Preact
<input onInput={e => console.log(e.target.value)} />
Run in REPL

If you're using preact/compat, most onChange events are internally converted to onInput to emulate React's behavior. This is one of the tricks we use to ensure maximum compatibility with the React ecosystem.

JSX Constructor

JSX is a syntax extension for JavaScript that is converted to nested function calls. The idea of using these nested calls to build up tree structures long predates JSX, and was previously popularized in JavaScript by the hyperscript project. This approach has value well beyond the scope of the React ecosystem, so Preact promotes the original generalized community-standard. For a more in-depth discussion of how JSX works and its relationship to Hyperscript, read this article.

Source: (JSX)

<a href="/">
  <span>Home</span>
</a>
Run in REPL

Output:

// Preact:
h(
  'a',
  { href:'/' },
  h('span', null, 'Home')
);

// React:
React.createElement(
  'a',
  { href:'/' },
  React.createElement('span', null, 'Home')
);
Run in REPL

Ultimately, if you're looking at the generated output code for a Preact application, it's clear that a shorter un-namespaced "JSX pragma" is both easier to read and more suitable for optimizations like minification. In most Preact apps you'll encounter h(), though it doesn't really matter which name you use since a createElement alias export is also provided.

No contextTypes needed

The legacy Context API requires Components to declare specific properties using React's contextTypes or childContextTypes in order to receive those values. Preact does not have this requirement: all Components receive all context properties produced by getChildContext() by default.

Features exclusive to preact/compat

preact/compat is our compatibility layer that translates React code to Preact. For existing React users this can be an easy way to try out Preact without changing any of your code, by setting up a few aliases in your bundler configuration.

Children API

The Children API is a specialized set of methods for working with the value of props.children. For Preact this is generally unnecessary, and we recommend using the built-in array methods instead. In Preact, props.children is either a Virtual DOM node, an empty value like null, or an Array of Virtual DOM nodes. The first two cases are the simplest and most common, since it's possible to use or return children as-is:

// React:
function App(props) {
  return <Modal content={Children.only(props.children)} />
}

// Preact: use props.children directly:
function App(props) {
  return <Modal content={props.children} />
}
Run in REPL

For specialized cases where you need to iterate over the children passed to a component, Preact provides a toChildArray() method that accepts any props.children value and returns a flattened and normalized Array of Virtual DOM nodes.

// React
function App(props) {
  const cols = Children.count(props.children);
  return <div data-columns={cols}>{props.children}</div>
}

// Preact
function App(props) {
  const cols = toChildArray(props.children).length;
  return <div data-columns={cols}>{props.children}</div>
}
Run in REPL

A React-compatible Children API is available from preact/compat to make integration with existing component libraries seamless.

Specialised Components

preact/compat ships with specialised components that are not necessary for every app. These include

  • PureComponent: Only updates if props or state have changed
  • memo: Similar in spirit to PureComponent but allows to use a custom comparison function
  • forwardRef: Supply a ref to a specified child component.
  • Portals: Continues rendering the current tree into a different DOM container
  • Suspense: experimental Allows to display fallback content in case the tree is not ready
  • lazy: experimental Lazy load async code and mark a tree as ready/not ready accordingly.