Skip to content

README.md

Latest commit

 

History

History
59 lines (42 loc) · 1.78 KB

README.md

File metadata and controls

59 lines (42 loc) · 1.78 KB

xsd-parser Benchmarks

This crate contains end-to-end performance and memory benchmarks for xsd-parser and the generated types.

It is intended for:

  • comparing different generated backends (e.g. quick-xml vs serde-based implementations),
  • tracking performance and memory regressions,

The goal is to benchmark realistic end-to-end flows.

Quickstart

Build and run in release mode:

cargo run -p benchmark --release

You can also pass CLI flags:

cargo run -p benchmark --release -- --help

If you are interested in benchmarks of the debug build you can also run the following command. This might be interesting because if you execute xsd-parser inside the build script of your crate, it is usually compiled using debug build.

cargo run -p benchmark

Outputs

The benchmark runner prints a summary table to the terminal

Typical metrics include:

  • Runtime (min / max / avg / median across runs),
  • Stack usage.

The reported stack usage is a high-water-mark estimate based on stack painting and scanning. It is Linux-only and should be treated as an approximation.

Adding a new benchmark case

  1. Extend the build script to generate the code from the schema
  2. Add a new module in benchmark/src/schemas/
  3. Register it in benchmark/src/schemas/mod.rs
  4. Implement the benchmark entrypoints used by the runner (benchmark/src/main.rs).

Guidelines:

  • Prefer end-to-end benchmarks that reflect real-world usage:
    • parse XML from bytes,
    • deserialize into the generated structs,
    • (optionally) re-serialize to XML.
  • Ensure test cases are reproducible:
    • fixed seeds,
    • fixed input payloads,
    • avoid reading from the network.
  • Use --release when comparing results.
  • Consider multiple payload sizes (small/medium/large) to get throughput curves.