Run npm run benchmark after npm run build:dist.
The harness reports wall-clock timings for plain reads, view creation and reads, arrays, Map, Set, Date, shallow freeze, and recursive deep freeze. It performs correctness checks before reporting and prints the Node/OS environment. Results are diagnostic rather than CI performance thresholds because shared runners are noisy. npm run benchmark:smoke verifies that every scenario executes.
Measured on Node 24.20.0, macOS arm64, 100,000 iterations:
| Scenario | Time |
|---|---|
| Plain nested read | 0.44 ms |
| Repeated view read | 15.38 ms |
| View creation | 26.20 ms |
| First nested access | 77.04 ms |
| Large object creation | 25.83 ms |
| Deep nested read | 23.52 ms |
| Array iteration (100 items) | 1,484.39 ms |
| Map read | 20.69 ms |
| Set iteration | 40.86 ms |
| Date read | 15.41 ms |
| Shallow freeze | 3.14 ms |
| Recursive deep freeze | 24.79 ms |
These numbers describe one local run and are not performance guarantees. The final reviewed alpha package baseline is 5,807 B minified ESM, 1,900 B minified+gzip, 2,026 B declarations, and a 6,521 B npm tarball. Enforced limits include maintenance headroom and live in scripts/size-budget.json.