perf-marks

Perf-marks

Why perf-marks?

That's the simplest and lightweight solution for User Timing API in Javascript. Simple how it should be.

You can find more details about it in the slides "User Timing API: because performance matters"

Contributing

Please check our contributing.md to know more about setup and how to contribute.

Setup and installation

Make sure that you are using the NodeJS version is the same as .nvmrc file version. If you don't have this version please use a version manager such as nvm or n to manage your local nodejs versions.

Please make sure that you are using NodeJS version 6.10.2

Assuming that you are using nvm, please run the commands inside this folder:

$ nvm install $(cat .nvmrc); # install required nodejs version
$ nvm use $(cat .nvmrc); # use nodejs version

In Windows, please install NodeJS using one of these options:

Via NVM Windows package: Dowload via this link. After that, run the commands:

$ nvm install $(cat .nvmrc); # install required nodejs version
$ nvm use $(cat .nvmrc); # use nodejs version

Via Chocolatey:

$ choco install nodejs.install -version 6.10.2

Install yarn

We use yarn as our package manager instead of npm

Install it following these steps

After that, just navigate to your local repository and run

$ yarn install

Demo

Try out our demo on Stackblitz!

Run the tests

$ yarn test # run the tests

Run the build

$ yarn build # run the tests

Run the bundlesize check

$ yarn bundlesize # run the tests

Run the code lint

$ yarn lint # run the tests

PerfMarks

This service exposes a few different methods with which you can interact with feature toggle service.

PerfMarks.start(markName)

Adds the user timing api marker instrumentation in your application.

import * as PerfMarks from 'perf-marks';

...
PerfMarks.start('name-of-your-mark');
...

PerfMarks.end(markName)

Returns the results for the specified marker.

PerfMarks.end(markName) calls PerfMarks.clear(markName) after return the mark values

import * as PerfMarks from 'perf-marks';

...
PerfMarks.start('name-of-your-mark');
...
const markResults: PerfMarks.PerfMarksPerformanceEntry = PerfMarks.end('name-of-your-mark');

PerfMarks.clear(markName)

Removes the specified marker

import * as PerfMarks from 'perf-marks';

...
PerfMarks.start('name-of-your-mark');
...
PerfMarks.clear('name-of-your-mark');
...

PerfMarks.clearAll()

Removes all the marker

import * as PerfMarks from 'perf-marks';

...
PerfMarks.start('name-of-your-mark');
PerfMarks.start('another-name-of-your-mark');
...
PerfMarks.clearAll();
...

PerfMarks.getNavigationMarker()

Gets the marks for navigation loaded

import * as PerfMarks from 'perf-marks';

...
const markResults: PerfMarksPerformanceNavigationTiming = PerfMarks.getNavigationMarker();
...

PerfMarks.getEntriesByType(markName)

Gets the result for all marks that matches with the given mark name

import * as PerfMarks from 'perf-marks';

...
PerfMarks.start('name-of-your-mark');
PerfMarks.start('another-name-of-your-mark');
...
// It will return results for all the marks that matches with `name-of-your-mark`
// In this case, `name-of-your-mark` and `another-name-of-your-mark`
const markResult: PerfMarksPerformanceNavigationTiming[] = PerfMarks.getEntriesByType('name-of-your-mark');
...

PerfMarks.isUserTimingAPISupported

Boolean with the result of the check if User Timing API is supported for the current browser/NodeJS version.

PerfMarks already have a fallback in case user timing is not supported. This boolean is exposed in case the app needs to check the case to use any other mechanism.

import * as PerfMarks from 'perf-marks';

...
if (PerfMarks.isUserTimingAPISupported) {
  // ... Do something
}
...

PerfMarks.isPerformanceObservableSupported

Boolean with the result of the check if PerformanceObservable is supported for the current browser/NodeJS version.

PerfMarks does not provide a fallback if PerformanceObservable is not supported. This boolean is exposed in case the app needs to check the case to use any other mechanism.

import * as PerfMarks from 'perf-marks';

...
// Checking if `PerformanceObservable` is supported for the current browser/NodeJS version
if (PerfMarks.isPerformanceObservableSupported) {
  try {
  // If yes, start the PerformanceObserver
    const observer: PerformanceObserver = new PerformanceObserver(list => {
      // ... Do something
    });

    // register observer based on the entryTypes
    // E.G. for long task notifications
    observer.observe({ entryTypes: ['longtask'] });
  } catch (e) {}
  // ... Finishing the observer
  observer.disconnect();
}
...

PerfMarks.profiler()

Runs profiler using User Timing Api method to get user timing information. It will return a Promise with mark key with a PerfMarksPerformanceEntry type OR mark key + data key with the content for the callback method If the given callback returns something.

const methodToBeMeasured = () => {
  /** method content */
};
// `res` will contain `mark` with the information and `data`
// if `methodToBeMeasured` returns something
const { mark, data } = PerfMarks.profiler(methodToBeMeasured, 'name-of-the-mark-for-this-method');

Entrypoints

These are entrypoints for specific components to be used carefully by the consumers. If you're using one of these entrypoints we are assuming you know what you are doing. So it means that code-splitting and tree-shaking should be done on the consumer/product side.

By definition it will use CJS as the main distribution entrypoint used in the app. However, this can be changed in the consumer's bundle step. This is the built-in scenario if the consumer uses toolings such as Webpack, Rollup, or Parcel.

Exposed entrypoints

• perf-marks/marks: it has all the methods for marks
  • start: Frontend and Backend support
  • end: Frontend and Backend support
  • clear: Frontend and Backend support
  • clearAll: Frontend and Backend support
  • isUserTimingAPISupported: Deprecated (will be removed in v2). Use the value imported from perf-marks/utils instead. Frontend and Backend support
  • isPerformanceObservableSupported: Deprecated (will be removed in v2). Use the value imported from perf-marks/utils instead. Frontend and Backend support
• perf-marks/entries: it has all the methods to get entries
  • getNavigationMarker: Frontend support only
  • getEntriesByType: frontend support only
• perf-marks/utils: it has all the feature, and platform checks and validations
  • isNodeJSEnv: Frontend and Backend support. Boolean with the result of the check if package is running on the browser or in a NodeJS environment
  • isPerformanceObservableSupported: Frontend and Backend support. Boolean with the result of the check if PerformanceObservable is supported for the current browser/NodeJS version
  • isUserTimingAPISupported: Frontend and Backend support. Boolean with the result of the check if User Timing API is supported for the current browser/NodeJS version
• perf-marks/profiler: it has all the feature, and platform checks and validations
  • profiler: Frontend and Backend support. profiler using User Timing Api method. It will return a Promise with mark key with a PerfMarksPerformanceEntry type or mark key + data key with the content for the callback method If the given callback returns something.

If you need optimize your bundle size even more, this package provides different bundles for CommonJS, UMD, ESM and ES2015. To make the dev experience smoothiest as possible, you can use babel-plugin-transform-imports in your app and configure the bundle that fits the most for your app!

yarn add -D babel-plugin-transform-imports
# or
npm install --save-dev babel-plugin-transform-imports

Create a .babelrc.js file in the root directory of your project:

const plugins = [
  [
    'babel-plugin-transform-imports',
    {
      'perf-marks/perf-marks': {
        // Use "transform: 'perf-marks/perf-marks/${member}'," if your bundler does not support ES modules
        transform: 'perf-marks/dist/esm/${member}',
        preventFullImport: true,
      },
      'perf-marks/entries': {
        // Use "transform: 'perf-marks/entries/${member}'," if your bundler does not support ES modules
        transform: 'perf-marks/entries/esm/${member}',
        preventFullImport: true,
      },
    },
  ],
];

module.exports = { plugins };

Or just use it via babel-plugin-import

yarn add -D babel-plugin-import
# or
npm install --save-dev babel-plugin-import

Create a .babelrc.js file in the root directory of your project:

const plugins = [
  [
    'babel-plugin-import',
    {
      libraryName: 'perf-marks/entries',
      // Use "'libraryDirectory': ''," if your bundler does not support ES modules
      libraryDirectory: 'esm',
      camel2DashComponentName: false,
    },
    'entries',
  ],
];

module.exports = { plugins };

And enjoy! Yeah, it's simple like that 😉

Publish

this project is using np package to publish, which makes things straightforward. EX: np <patch|minor|major>

For more details, please check np package on npmjs.com

Author

Wilson Mendes (willmendesneto)

• https://twitter.com/willmendesneto
• http://github.com/willmendesneto

Changelog

[1.12.0][] - 2020-07-31

Added

• Adding entry points checks in bundlesize task
• New method profiler() to profile functions using perf-marks. The idea here is to avoid boilerplates on consumer's side by having an easy way to do that. So instead of doing this

import { start, end } from 'perf-marks/marks';
const markName = 'name-of-the-mark-for-this-method';

const methodToBeMeasured = () => {
  /** method content */
};
// ...
start(markName);
const data = methodToBeMeasured();
const mark = end(markName);
return { data, mark };

The consumers can have same effect by doing this

const methodToBeMeasured = () => {
  /** method content */
};
// `res` will contain `mark` with the information and `data`
// if `methodToBeMeasured` returns something
const res = profiler(methodToBeMeasured, 'name-of-the-mark-for-this-method');

Fixed

• Fixing entry point for perf-marks/utils