React Native Performance API

This is an implementation of the Performance API for React Native based on the User Timing Level 3 and Performance Timeline Level 2 drafts.

Note: The timestamps used are high resolution (fractions of milliseconds) and monotonically increasing, meaning that they are independent of system clock adjustments. To convert a performance timestamp to a unix epoch timestamp do like this:

const timestamp = Date.now() - performance.timeOrigin + entry.startTime;

Installation

Yarn: yarn add --dev react-native-performance

NPM: npm install --save-dev react-native-performance

Manual integration

If your project is not set up with autolinking you need to link manually.

iOS

Add the following to your Podfile and run pod install:

pod 'react-native-performance', :path => '../node_modules/react-native-performance/ios'

Usage

See examples/vanilla for a demo of the different features.

Basic measure example

Marking timeline events, measuring the duration between them and fetching these entries works just like on the web:

import performance from 'react-native-performance';

performance.mark('myMark');
performance.measure('myMeasure', 'myMark');
performance.getEntriesByName('myMeasure');
-> [{ name: "myMeasure", entryType: "measure", startTime: 98, duration: 123 }]

Meta data

If you want to add some additional details to your measurements or marks, you may pass a second options object argument with a detail entry per the User Timing Level 3 draft:

import performance from 'react-native-performance';

performance.mark('myMark', {
  detail: {
    screen: 'settings',
    ...
  }
});
performance.measure('myMeasure', {
  start: 'myMark',
  detail: {
    category: 'render',
    ...
  }
});
performance.getEntriesByType('measure');
-> [{ name: "myMeasure", entryType: "measure", startTime: 98, duration: 123, detail: { ... } }]

Subscribing to entries

The PerformanceObserver API enables subscribing to different types of performance entries. The handler is called in batches.

Passing buffered: true would include entries produced before the observe() call which is useful to delay handing of measurements until after performance critical startup processing.

import { PerformanceObserver } from 'react-native-performance';
const measureObserver = new PerformanceObserver((list, observer) => {
  list.getEntries().forEach((entry) => {
    console.log(`${entry.name} took ${entry.duration}ms`);
  });
});
measureObserver.observe({ type: 'measure', buffered: true });

Network resources

Resource logging is disabled by default and currently will only cover fetch/XMLHttpRequest uses.

import performance, {
  setResourceLoggingEnabled,
} from 'react-native-performance';

setResourceLoggingEnabled(true);

await fetch('https://domain.com');
performance.getEntriesByType('resource');
-> [{
  name: "https://domain.com",
  entryType: "resource",
  startTime: 98,
  duration: 123,
  initiatorType: "xmlhttprequest", // fetch is a polyfill on top of XHR in react-native
  fetchStart: 98,
  responseEnd: 221,
  transferSize: 456,
  ...
}]

Custom metrics

If you want to collect custom metrics not based on time, this module provides an extension of the Performance API called .metric() that produces entries with the type metric.

import performance from 'react-native-performance';

performance.metric('myMetric', 123);
performance.getEntriesByType('metric');
-> [{ name: "myMetric", entryType: "metric", startTime: 98, duration: 0, value: 123 }]

Native marks

This library exposes a set of native timeline events and metrics such as native app startup time, script execution time etc under the entryType react-native-mark.

To install the native iOS dependency required, simply run pod install in ios/ directory and rebuild the project. For android it should be enough by just rebuilding.

If you wish to opt out of autolinking of the native dependency, you may create or alter the react-native.config.js file to look something like this:

// react-native.config.js

module.exports = {
  dependencies: {
    'react-native-performance': {
      platforms: {
        android: null,
        ios: null,
      },
    },
  },
};

Note that the native marks are not available immediately upon creation of the JS context, so it's best to set up an observer for the relevant end event before making measurements.

import performance, { PerformanceObserver } from 'react-native-performance';

new PerformanceObserver((list, observer) => {
  if (list.getEntries().find((entry) => entry.name === 'runJsBundleEnd')) {
    performance.measure('nativeLaunch', 'nativeLaunchStart', 'nativeLaunchEnd');
    performance.measure('runJsBundle', 'runJsBundleStart', 'runJsBundleEnd');
  }
}).observe({ type: 'react-native-mark', buffered: true });

Custom marks

ephemeral is an optional parameter to mark/metric functions which if set to NO/false will retain the entries when the React Native bridge is (re)loaded.

iOS

#import <react-native-performance/RNPerformance.h>

[RNPerformance.sharedInstance mark:@"myCustomMark"];
[RNPerformance.sharedInstance mark:@"myCustomMark" detail:@{ @"extra": @"info" }];
[RNPerformance.sharedInstance mark:@"myCustomMark" ephemeral:NO];

[RNPerformance.sharedInstance metric:@"myCustomMetric" value:@(123)];
[RNPerformance.sharedInstance metric:@"myCustomMetric" value:@(123) detail:@{ @"unit": @"ms" }];
[RNPerformance.sharedInstance metric:@"myCustomMetric" value:@(123) ephemeral:NO];

Android

import com.oblador.performance.RNPerformance;

RNPerformance.getInstance().mark("myCustomMark");
RNPerformance.getInstance().mark("myCustomMark", false); // ephermal flag to disable resetOnReload
Bundle bundle = new Bundle();
bundle.putString("extra", "info");
RNPerformance.getInstance().mark("myCustomMark", bundle); // Bundle to pass some detail payload

RNPerformance.getInstance().metric("myCustomMetric", 123);
RNPerformance.getInstance().metric("myCustomMetric", 123, false); // ephermal flag to disable resetOnReload
Bundle bundle = new Bundle();
bundle.putString("unit", "ms");
RNPerformance.getInstance().metric("myCustomMetric", 123, bundle); // Bundle to pass some detail payload

Supported marks

Name	Platforms	Description
`nativeLaunchStart`	Both	Native process initialization started
`nativeLaunchEnd`	Both	Native process initialization ended
`downloadStart`	Both	Only available in development. Development bundle download started
`downloadEnd`	Both	Only available in development. Development bundle download ended
`runJsBundleStart`	Both	Not available with debugger. Parse and execution of the bundle started.
`runJsBundleEnd`	Both	Not available with debugger. Parse and execution of the bundle ended
`contentAppeared`	Both	Initial component mounted and presented to the user.
`bridgeSetupStart`	Both
`bridgeSetupEnd`	iOS
`reactContextThreadStart`	Android
`reactContextThreadEnd`	Android
`vmInit`	Android
`createReactContextStart`	Android
`processCoreReactPackageStart`	Android
`processCoreReactPackageEnd`	Android
`buildNativeModuleRegistryStart`	Android
`buildNativeModuleRegistryEnd`	Android
`createCatalystInstanceStart`	Android
`createCatalystInstanceEnd`	Android
`preRunJsBundleStart`	Android
`createReactContextEnd`	Android
`preSetupReactContextStart`	Android
`preSetupReactContextEnd`	Android
`setupReactContextStart`	Android
`attachMeasuredRootViewsStart`	Android
`createUiManagerModuleStart`	Android
`createViewManagersStart`	Android
`createViewManagersEnd`	Android
`createUiManagerModuleConstantsStart`	Android
`createUiManagerModuleConstantsEnd`	Android
`createUiManagerModuleEnd`	Android
`attachMeasuredRootViewsEnd`	Android
`setupReactContextEnd`	Android