main

Warming up the performance wizard...

Python/uv project, pytest; no existing benchmarks

Analyze repository

Add pytest-codspeed benchmarks

Local simulation run uploaded successfully

Run benchmarks locally with CodSpeed

Add CI workflow and badge

PR #5160: all CI green, CodSpeed comment posted, marked ready

Open PR and monitor CI

Benchmarks have been successfully run and processed. The results are available in CodSpeed

Processing benchmark results

CodSpeed is set up for litestar in PR https://github.com/litestar-org/litestar/pull/5160. All CI checks passed, and the CodSpeed bot posted its installation comment showing 21 new benchmarks. The PR is marked ready and Kumzy is requested as reviewer.

**What the PR adds:**
- **Dependency:** `pytest-codspeed` in the `test` dependency group, with `uv.lock` updated.
- **21 benchmarks in `tests/benchmarks/`**, measured with CodSpeed's CPU simulation mode:
  - **Request handling (12):** requests go straight into the app in-process rather than through `TestClient`, so the numbers reflect framework work. They cover:
    - plaintext responses (sync and async handlers) and JSON responses
    - path, query and header parameters
    - JSON bodies (msgspec struct and dict)
    - a DTO round trip and a dependency chain
    - 404s and validation errors
  - **App level (6):** startup for a small and a large app (120 routes), OpenAPI schema generation, and route lookup in the routing table (the lookup cache is bypassed so the lookup itself is measured).
  - **Serialization (3):** JSON encoding and decoding.
- **CI workflow:** `.github/workflows/codspeed.yml` runs the benchmarks on pushes to `main`, on pull requests and on manual dispatch. It uses `CodSpeedHQ/action@v5.6.0` with OIDC authentication.
- **Badge:** a CodSpeed badge in the README. A pre-commit hook copied the same change into `docs/PYPI_README.md`.

The benchmarks are ordinary pytest tests, so without `--codspeed` they also run once as part of the normal test suite.

**Next step:** merge the PR so `main` gets a baseline. After that, pull requests will get performance reports.

Performance History

Latest Results

Latest Branches