Benchmarks

This directory contains performance benchmarks for node-casbin.

Running Benchmarks Locally

To run the benchmarks locally:

yarn benchmark

This will:

1. Build the CommonJS version of the library
2. Build the benchmark suite
3. Run all benchmarks and display results

What is Benchmarked

The benchmark suite tests the performance of:

RBAC Model

• enforce() (async) - both allow and deny cases
• enforceSync() - both allow and deny cases
• getRolesForUser() - get user roles
• hasRoleForUser() - check if user has a specific role

ABAC Model

• enforce() (async) - attribute-based access control
• enforceSync() - attribute-based access control

Basic Model

• enforce() (async) - basic access control
• enforceSync() - basic access control

Policy Management

• getPolicy() - retrieve all policies
• hasPolicy() - check if policy exists
• getFilteredPolicy() - retrieve filtered policies

Automated Benchmarking

The benchmark workflow automatically runs on every Pull Request:

1. Runs benchmarks on PR branch - measures performance of proposed changes
2. Runs benchmarks on base branch - establishes baseline performance
3. Compares results - calculates percentage changes
4. Posts comment to PR - displays results in an easy-to-read table

Understanding Benchmark Results

The PR comment will show:

• 🚀 - Significant improvement (>5%)
• ✅ - Improvement (0-5%)
• ➖ - No significant change
• ⬇️ - Minor regression (0-5%)
• ⚠️ - Regression (>5%)

What to Do About Regressions

If your PR shows performance regressions:

1. Review the changes - identify what might cause the slowdown
2. Profile the code - use Node.js profiling tools to find bottlenecks
3. Consider alternatives - can the same functionality be achieved more efficiently?
4. Document trade-offs - if the regression is unavoidable, document why the change is necessary

Small regressions (<5%) are generally acceptable if:

• The change adds important functionality
• The change improves code maintainability
• The change fixes a bug or security issue

Adding New Benchmarks

To add new benchmarks, edit benchmark/benchmark.ts:

// Create a new suite
const mySuite = createSuite('My Feature');

mySuite
  .add('My benchmark', {
    defer: true, // for async tests
    fn: async (deferred: Benchmark.Deferred) => {
      await myFunction();
      deferred.resolve();
    },
  })
  .on('complete', () => {
    resolve();
  })
  .run({ async: true });

For synchronous tests, omit the defer option:

mySuite.add('My sync benchmark', () => {
  mySyncFunction();
});