torchcell.knowledge_graphs.kg_manifest.check_admission

check_admission(manifest: KgBuildManifest, repo_root: Path, dataset_class_name: str, data_root: Path, ack_adapter_drift: str | None = None, ack_value_drift: str | None = None, served_experiment_ids: Callable[[str], Iterable[str]] | None = None, served_source: str = 'live store', include_private: bool = False, widen_organisms: bool = False) → AdmissionReport

Decide whether dataset_class_name can be added to the store incrementally.

A dataset whose declared organism is outside the served organism set (served_organisms()) blocks unless widen_organisms (admit --widen-organisms): a yeast-only store does not quietly become a mixed one. With the flag the report says the admission widens the set, and record writes it.

served_experiment_ids reads the experiment ids the live store holds under a Dataset node (live_experiment_ids bound to a connection in production, named by served_source in the report). It is called only when the dataset is already served, to run the superset proof; without it a served dataset blocks.

A PRIVATE dataset blocks unless include_private (admit --include-private), the same opt-in the build takes: the in-house store carries private data, a public one never does. With the flag the private adapter map is searched too.