Benchmarks

Performance baseline for core flows. Regressions can be checked in CI with a timing gate (see Phase 9).

Command

pytest capper/tests/benchmark_core.py -v --benchmark-only

Benchmarks are marked with @pytest.mark.benchmark and are excluded from the default test run (pytest -m 'not benchmark').

Baseline (reference)

Run on a typical developer machine (Darwin, Python 3.11). Your numbers will vary.

Test	Mean (μs)	OPS (Kops/s)
ModelFactory.build() (Name, Email)	~237	~4.2
DataclassFactory.build() (Name, Email)	~166	~6.0
Hypothesis strategy.example() for Name	~347	~2.9

• ModelFactory: Pydantic model with 2 Capper types; one .build() per iteration.
• DataclassFactory: Dataclass with 2 Capper types; one .build() per iteration.
• Hypothesis: One draw from st.from_type(Name) via .example() per iteration.

CI uses a lightweight performance check (e.g. 1000× UserFactory.build() in under a threshold) rather than comparing to this baseline; this document is for manual regression checks and reference.