ember-batcher

ember-batcher

An Ember addon to batch DOM reads and mutations using requestAnimationFrame.

As noted in Paul Irish's "What Forces Layout", a number of properties or methods, when requested/called, will trigger the browser to sychronously calculate the style and layout. This is also called reflow or layout thrashing, and is common performance bottleneck. This addon attempts to provide a mechanism to compliment Ember's runloop batching to help minimize layout thrashing.

Compatibility

• Ember.js v3.24 or above
• Ember CLI v3.24 or above
• Node.js v14 or above

Installation

ember install ember-batcher

Usage

readDOM(readTask: Function): void

Register a task function that will get batched with other "reads" and called on the next requestAnimationFrame (if supported). The method will be executed within either the current run loop or will create a new run loop if necessary.

import Component from '@glimmer/component';
import { readDOM } from 'ember-batcher';

export default class MyComponent extends Component {
  foo() {
    readDOM(() => {
      // Perform DOM read
    });
  }
}

mutateDOM(mutationTask: Function): void

Register a task function that will get batched with other "mutations" and before other "reads". The method will be called on the next requestAnimationFrame (if supported). The method will be executed within either the current run loop or will create a new run loop if necessary.

import Component from '@glimmer/component';
import { mutateDOM } from 'ember-batcher';

export default class MyComponent extends Component {
  foo() {
    mutateDOM(() => {
      // Perform DOM mutation
    });
  },
}

Reads, then Writes

Since the purpose of this addon is to minimize layout thrashing, we want to perform all reads before we perform writes. Additionally, you should close over values created during the read phase that are used in the write phase. This helps ensure that work doesn't leak outside of the frame we're performing the work in, and allows you to prepare values in the read phase that are used in the subsequent write phase.

To be clear, writes should not occur during the reads phase, therefore you should not have any code that performs any reads from within the callback of readDOM. All writes should occur inside a call to mutateDOM, whether that's in a non-nested call to _mutateDOM itself, or from a nested call to mutateDOM inside of a readDOM call.

Example of read first, then write.

import Component from '@glimmer/component';
import { tracked } from '@glimmer/tracking';
import { readDOM, mutateDOM } from 'ember-batcher';

export default class MyComponent extends Component {
  @tracked
  width;

  foo() {
    readDOM(() => {
      const width = document.querySelector('.foo').clientWidth;

      mutateDOM(() => {
        // we should perform our update conditionally, only if the value of width changes. This helps
        // minimize unnecessary layout thrashing.
        if (this.width !== width) {
          this.width = width * 2;
        }
      });
    });
  },
}

Contributing

See the Contributing guide for details.

License

This project is licensed under the MIT License.

Changelog

v5.0.0 (2022-11-17)

:boom: Breaking Change

• #325 Update to support Ember 3.28 and 4.4 (@mikrostew)
• #222 Updating ember-try versions (@scalvert)
• #217 Drop support for Node 8 (@scalvert)

Committers: 3

• Michael Stewart (@mikrostew)
• Steve Calvert (@scalvert)
• Travis Hoover (@thoov)