Skip to main content

Get Started

Step 1: Install the SDK​

# or any package manager you are using, e.g. npm/pnpm
yarn add @perfsee/package -D

Step 2: Create Benchmark files (optional)​

Create a file named test.bench.js under the project.

const Benchmark = require('@perfsee/package')

Benchmark('foo', () => {
  bar()
})

Benchmark('bar', [
  {
    test: () => {
      baz()
    },
    options: { name: 'baz' },
  },
  {
    test: async () => {
      await setup()
      return () => {
        quz()
      }
    },
    options: { name: 'quz with async setup' },
  },
])

More about benchmark api, please refer to Benchmark API.

Step 3: Analyze and upload (ONLY on CI)​

Before analysis, it's necessary to create a perfsee project and apply API Token on the Token Management page.

PERFSEE_TOKEN=<your-token> npx @perfsee/package <path-to-package> --project=<perfsee-project-id>

If there are benchmark files named xxx.{bench,benchmark}.{js,ts}(by default), this command will run benchmarks before uploading.

On CI environment, the result will be uploaded to the platform. Otherwise, results will be shown in a html page.

Cli Options​

project​

Id of your perfsee project on the platform.

customImports​

By default, the default export is used for calculating sizes. Setting this option allows calculation of package stats based on more granular top-level exports.

minifier: 'esbuild' | 'terser'​

default: 'esbuild'​

ESbuild is faster, albeit with marginally larger file sizes.

target: 'browser' | 'node'​

default: 'node'​

If target is browser, benchmarks will be bundled and run in headless chrome (on server if in CI environment).

benchmarkPattern​

default: '*.{bench|benchmark}.{js|ts}'​

Glob pattern to find benchmark files.

benchmarkTimeout​

Timeout for benchmark running. Unit: sec.

Step 4: View the report​

When the analysis finished, we will automatically generate a report link. It will show the result of your package after a while. For more details, please refer to Report Details.