node package manager
Painless code sharing. npm Orgs help your team discover, share, and reuse code. Create a free org »


nanocomponent stability

npm version build status downloads js-standard-style

Native DOM components that pair nicely with DOM diffing algorithms.


  • Isolate native DOM libraries from DOM diffing algorithms
  • Makes rendering elements very fast™ by avoiding unnecessary rendering
  • Component nesting and state update passthrough
  • Implemented in only a few lines
  • Only uses native DOM methods
  • Class based components offering a familiar component structure
  • Works well with bel and yoyoify
  • Combines the best of nanocomponent@5 and cache-component@5.


// button.js
var Nanocomponent = require('nanocomponent')
var html = require('bel')
function Button () {
  if (!(this instanceof Button)) return new Button()
  this.color = null
Button.prototype = Object.create(Nanocomponent.prototype)
Button.prototype.createElement = function (color) {
  this.color = color
  return html`
    <button style="background-color: ${color}">
      Click Me
// Implement conditional rendering
Button.prototype.update = function (newColor) {
  return newColor !== this.color
// index.js
var choo = require('choo')
var Button = require('./button.js')
var button = Button()
var app = choo()
app.route('/', mainView)
function mainView (state, emit) {
  return html`
app.use(function (state, emitter) {
  state.color = 'green'


These are some common patterns you might encounter when writing components.


Nanocomponent is part of the choo ecosystem, but works great standalone!

var Button = require('./button.js')
var button = new Button()
// Attach to DOM
// Update mounted component
// Log a reference to the mounted dom node

Binding event handlers as component methods

Sometimes it's useful to be pass around prototype methods into other functions. This can be done by binding the method that's going to be passed around:

var Nanocomponent = require('nanocomponent')
var html = require('bel')
function Component () {
  if (!(this instanceof Component)) return new Component()
  // Bind the method so it can be passed around
  this.handleClick = this.handleClick.bind(this)
Component.prototype = Object.create(Nanocomponent.prototype)
Component.prototype.handleClick = function (ev) {
  console.log('element is', this.element)
Component.prototype.createElement = function () {
  return html`<button onClick=${this.handleClick}>
    My component
Component.prototype.update = function () {
  return false // Never re-render

ES6 Class Syntax

Because Class syntax is just sugar for prototype code, Nanocomponent can be written using Classes too:

var Nanocomponent = require('nanocomponent')
var html = require('bel')
class Component extends Nanocomponent {
  constructor () {
    this.color = null
  createElement (color) {
    this.color = color
    return html`
      <div style="background-color: ${color}">
        Color is ${color}
  update (newColor) {
    return newColor !== this.color

Mutating the components instead of re-rendering

Sometimes you might want to mutate the element that's currently mounted, rather than performing DOM diffing. Think cases like third party widgets that manage themselves.

var Nanocomponent = require('nanocomponent')
var html = require('bel')
function Component () {
  if (!(this instanceof Component)) return new Component()
  this.text = ''
Component.prototype = Object.create(Nanocomponent.prototype)
Component.prototype.createElement = function (text) {
  this.text = text
  return html`<h1>${text}</h1>`
Component.prototype.update = function (text) {
  if (text !== this.text) {
    this.text = text
    this.element.innerText = this.text   // Directly update the element
  return false                           // Don't call createElement again
Component.prototype.unload = function (text) {
  console.log('No longer mounted on the DOM!')

Nested components and component containers

Components nest and can skip renders at intermediary levels. Components can also act as containers that shape app data flowing into view specific components.

var Nanocomponent = require('nanocomponent')
var html = require('bel')
var Button = require('./button.js')
function Component () {
  if (!(this instanceof Component)) return new Component()
  this.button1 = new Button ()
  this.button2 = new Button ()
  this.button3 = new Button ()
Component.prototype = Object.create(Nanocomponent.prototype)
Component.prototype.createElement = function (state) {
  var colorArray = shapeData(state)
  return html`
Component.prototype.update = function (state) {
  var colorArray = shapeData(state) // process app specific data in a container
  this.button1.render(colorArray[0]) // pass processed data to owned children components
  return false // always return false when mounted
// Some arbitrary data shaping function
function shapeData (state) {
  return [state.colors.color1, state.colors.color2, state.colors.color3]


What order do lifecycle events run in?

Lifecycle diagram Note: aftercreate should actually say afterupdate. Shoutout to @lrlna for the excellent diagram.

Where does this run?

Nanocomponent was written to work well with choo, but it also works well with DOM diffing engines that check .isSameNode() like nanomorph and morphdom. It is designed and documented in isolation however, so it also works well on it's own if you are careful. You can even embed it in other SPA frameworks like React or Preact with the use of nanocomponent-adapters which enable framework-free components! 😎

What's a proxy node?

It's a node that overloads Node.isSameNode() to compare it to another node. This is needed because a given DOM node can only exist in one DOM tree at the time, so we need a way to reference mounted nodes in the tree without actually using them. Hence the proxy pattern, and the recently added support for it in certain diffing engines:

var html = require('bel')
var el1 = html`<div>pink is the best</div>`
var el2 = html`<div>blue is the best</div>`
// let's proxy el1
var proxy = html`<div></div>`
proxy.isSameNode = function (targetNode) {
  return (targetNode === el1)
el1.isSameNode(el1)   // true
el1.isSameNode(el2)   // false
proxy.isSameNode(el1) // true
proxy.isSameNode(el2) // false

How does it work?

nanomorph is a diffing engine that diffs real DOM trees. It runs a series of checks between nodes to see if they should either be replaced, removed, updated or reordered. This is done using a series of property checks on the nodes.

nanomorph runs Node.isSameNode(otherNode) when diffing two DOM trees. This allows us to override the function and replace it with a custom function that proxies an existing node. Check out the code to see how it works. The result is that if every element in our tree uses nanocomponent, only elements that have changed will be recomputed and re-rendered making things very fast.

nanomorph, which saw first use in choo 5, has supported isSameNode since its conception. morphdom has supported .isSameNode since v2.1.0.

Is this basically react-create-class?

nanocomponent is very similar to react-create-class, but it leaves more decisions up to you. For example, there is no built in props or state abstraction in nanocomponent but you can do something similar with arguments (perhaps passing a single props object to .render e.g. .render({ foo, bar }) and assigning internal state to this however you want (perhaps this.state = { fizz: buzz }).


component = Nanocomponent([name])

Create a new Nanocomponent instance. Additional methods can be set on the prototype. Takes an optional name which is used when emitting timings.


Render the component. Returns a proxy node if already mounted on the DOM. Proxy nodes make it so DOM diffing algorithms leave the element alone when diffing. Call this when arguments have changed.


Re-run .render using the last arguments that were passed to the render call. Useful for triggering component renders if internal state has changed. Arguments are automatically cached under this._arguments (🖐 hands off, buster! 🖐). The update method is bypassed on re-render.


A getter property that returns the component's DOM node if its mounted in the page and null when its not.

DOMNode = Nanocomponent.prototype.createElement([arguments…])

Must be implemented. Component specific render function. Optionally cache argument values here. Run anything here that needs to run along side node rendering. Must return a DOMNode. Use beforerender to run code after createElement when the component is unmounted. Previously named _render. Arguments passed to render are passed to createElement. Elements returned from createElement must always return the same root node type.

Boolean = Nanocomponent.prototype.update([arguments…])

Must be implemented. Return a boolean to determine if prototype.createElement() should be called. The update method is analogous to React's shouldComponentUpdate. Called only when the component is mounted in the DOM tree. Arguments passed to render are passed to update.


A function called right after createElement returns with el, but before the fully rendered element is returned to the render caller. Run any first render hooks here. The load and unload hooks are added at this stage. Do not attempt to rerender in beforerender as the component may not be in the DOM yet.


Called when the component is mounted on the DOM. Uses on-load under the hood.


Called when the component is removed from the DOM. Uses on-load under the hood.


Called after a mounted component updates (e.g. update returns true). You can use this hook to call element.scrollIntoView or other dom methods on the mounted component.


Called after a component is re-ordered. This method is rarely needed, but is handy when you have a component that is sensitive to temorary removals from the DOM, such as externally controlled iframes or embeds (e.g. embedded tweets).


$ npm install nanocomponent

Optional lifecycle events

You can add even more lifecycle events to your components by attatching the following modules in the beforerender hook.

See also


Similar Packages