Examples
Each example is a complete, buildable project that walks through a real algorithm from scaffold to optimised implementation.
| Example | What it demonstrates |
|---|---|
| Running stats | Welford online mean/variance; introductory single-component walkthrough |
| FIR filter | Stateful step(), just-makeit perf, JM_DEFINE_STEPS |
| Sliding correlator | Complex cross-correlation, JM_DEFINE_STEPS with a different state layout |
| Sliding power | --return-type float, jm_simd.h recompute, numerical drift analysis |
| Array processing | All six array-processing patterns: auto steps(), methods, --variable-output, multi-output, --arg-type type[], --out-type |
| Stream chunker | Variable-size input and output: --no-step + --variable-output; accumulate-and-fire with irregular bursts |
| DSP toolkit | Multi-object library (Gain + EMA); just-makeit object, __init__.py auto-splice |
| Filter module | Multi-type module (Fir + Biquad in one .so); just-makeit module + just-makeit object |
| IQ file | cf32 ↔ q15 converter; --field properties, generator object, pip install -e ., wheel build |
All examples ship with end-to-end tests in examples/*/test.py that are run
by the CI suite. See examples/README.md for contributor notes on the
.steps/ naming convention.