nirs4all.pipeline.dagml.generated_subprocess module

Run generated-view DAG-ML campaigns in an isolated Python interpreter.

nirs4all.pipeline.dagml.generated_subprocess.run_generated_subprocess(*, dsl: dict[str, Any], envelope: dict[str, Any], graph: dict[str, Any], dataset: Any, dataset_path: str, dataset_pickle: str | None, workdir: Any, venv_python: str | None, selection_metric: str, sample_metadata: dict[str, dict[str, Any]] | None, random_state: int | None, refit: bool, refit_top_k: int) → dict[str, Any]

Transfer the PLAN provider and return the child’s native CV/refit outcome.

The live receipt store is never serialized. Only a trusted run-local request and response cross the process boundary; the child creates its own store.