@onekeyfe/react-native-perf-memory

Native process memory sampling for OneKey React Native applications on iOS and Android. The Nitro module returns one numeric byte count under the historical rss field. Callers own sampling frequency, aggregation and presentation; the package does not start a recurring sampler or upload measurements.

Installation

yarn add @onekeyfe/react-native-perf-memory react-native-nitro-modules

Install the application's iOS pods and rebuild the native application. The measurement is performed by native code and requires the Nitro runtime.

Usage

import { ReactNativePerfMemory } from '@onekeyfe/react-native-perf-memory';

const { rss } = await ReactNativePerfMemory.getMemoryUsage();
const memoryMiB = rss / (1024 * 1024);

Measurement semantics

getMemoryUsage() returns Promise<{ rss: number }> with the value in bytes. Despite the shared field name, the preferred measurements differ:

| Platform | Preferred sample | Fallback | | --- | --- | --- | | iOS | Mach phys_footprint | Mach resident size, then zero. | | Android | /proc/self/status VmRSS | ActivityManager total PSS converted from KiB, then zero. |

A zero result can indicate that a measurement could not be obtained. It is not proof that the application has no memory usage. The number covers native process memory rather than only the JavaScript heap, and should not be compared across platforms as if both use an identical accounting method.

Use bounded sampling intervals suited to diagnostics rather than requesting a sample on every render. The exported interface is defined in src/ReactNativePerfMemory.nitro.ts.

License

MIT.