Most of the Javascript frameworks like Vue, React and Ember implement a “virtual DOM”.
While it seems like some thing from a technological know-how fiction, its number one motive for present is to expand the speed and proficiency of DOM updates. It offers a few more advantages as well.
What the DOM actually is (a reminder)
We have a tendency consider the DOM as the HTML document it represents. but without a doubt the DOM is a tree-like data structure that comes into existence as soon as an HTML document has been parsed
The browser paints the DOM to the screen and will repaint it inresponse to user actions (e.g. mouse clicks) and updates through its API from your Javascript scripts e.g. document.createElement.
It’s expensive to update the DOM
When we use Javascript to make a change to our page, the browser has to do some work to find the required DOM nodes and make the change e.g.
// #myId could be anywhere in the document, which may have thousands of nodes! document.getElementById('myId').appendChild(myNewNode);
In modern apps there can be thousands of nodes in the DOM, and so updates can be computationally expensive. It’s inevitable that small, frequent updates will slow the page down.
What is a virtual DOM?
The DOM can be represented as a data structure in Javascript, too. Here is pseudo-code of ways a DOM node may be represented:
// An unordered list represented as Javascript let domNode = { tag: 'ul', attributes: { id: 'myId' }, children: [ // where the LI's would go ] };
If we call that a “virtual” DOM node, then the complete structure of DOM nodes would make up our virtual DOM.
But why do this?
It’s not very expensive to update virtual nodes.
// This might be how we update the virtual DOM domNode.children.push('
- Item 3
');
If we use a virtual DOM, in preference to our code directly calling the DOM API with methods like .getElementById to make updates, the code will make adjustments just to the JS object, which is cheap.
Then, whilst it’s time to get the real DOM in sync with the modifications we’ve made, an efficient updating function is used:
// This function would call the DOM API and make changes // to the browser's DOM. It would do it in batches and with // more efficiency than it would with arbitrary updates. sync(originalDomNode, domNode);
In any one cycle there may be many nodes to update, so batching API calls in this way could reduce a lot of inefficiency.
More than performance
Having a virtual DOM it not just a performance enhancement, it means additional functionality will be possible.
For example, in Vue.js, we can bypass the need for an HTML template or a template property by using a render() method, which returns virtual nodes:
new Vue({ el: '#app', data: { message: 'hello world' }, render(createElement) { return createElement( 'div', { attrs: { id: 'myId' } }, this.message ); } });
Output:
<div id='app'> <div id='myId'>hello world</div> </div>
Why do that? There are several viable benefits:
- You get the programmatic strength of Javascript. you could create factory-style functions to build your virtual nodes using Javascript’s array methods etc, something that might be more difficult using template syntax.
- You may make your code universal. since your Vue instance does not really on an HTML file it is also renderable by a server for server-side rendering.
- JSX. Render functions allow JS extensions like JSX which may be acceptable for architecting a component-based app.
Note: this article was originally posted here on the Vue.js Developers blog on 2017/02/21