Benchmarks

httpr's performance is tracked three ways. All of them run in GitHub Actions.

Comparison with other clients

Every Monday, benchmark/benchmark.py runs httpr against the latest release of requests, httpx, httpx2, aiohttp, curl_cffi, pycurl and tls_client: 400 requests per cell against a local uvicorn server with gzip responses, in sync, async and multi-threaded modes. Wall-clock seconds, lower is better.

The full tables and raw CSVs are published with the chart. Absolute numbers depend on the GitHub runner, so compare rows within a chart rather than charts across weeks.

httpr over time

On every push to main, tests/benchmark/test_performance.py runs under pytest-benchmark and the results are appended to the live trend chart. The runners are shared, so a single run can swing by 1.5x; look at the trend, not at individual points.

CPU cost per pull request

CodSpeed measures the CPU instructions of tests/benchmark/codspeed/ on every pull request and reports the difference against main on the PR. This is deterministic to about one percent and is what catches regressions before they merge.

Running locally

See benchmark/README.md.