ordinator

A dataset passed through named, swappable, recorded steps. This is the machinery: the plan and its execution, the step and engine registries, settings, provenance and trace, the bridge to another project's functions, and the web page that drives an engine. It has no vocabulary of its own. A consumer declares its words as data and answers what the kernel asks through the names it declares; onefluxpro is the first consumer.

Nothing is imported on import ordinator: take the modules you need.

Install

pip install ordinator            # the kernel
pip install "ordinator[bridge]"  # with numpy and pandas, for bridged libraries

From a checkout, pip install -e ".[test]" and python -m pytest. The tests run with no consumer installed, which is the point of them: what a consumer answers the kernel is tested where the consumer is.

Where it came from

Carved out of onefluxpro, its first consumer, on 2026-11-07, and given a repository of its own on 2026-11-13. The history here is that of src/ordinator and tests/kernel there. Its siblings are regorator, which registers, and locutor, which calls R and Julia; this one orders.