API reference

The hedbench package holds the shared benchmark tooling. Both modules are stubs until the standardized benchmark JSON format is designed (see the json_schemas/ directory in the repository).

hedbench.runner

Discover and run benchmark case studies through the standardized JSON interface.

Each case study under use_cases/ will declare itself with a JSON descriptor validated against the schemas in json_schemas/. Until that format is designed (see json_schemas/README.md), case studies are run directly by their own scripts.

hedbench.runner.run_case(case_dir, run_input=None)

Run one benchmark case study from its descriptor.

Parameters:

• case_dir – Path to the case-study directory under use_cases/.

• run_input – Optional path to a run-input JSON document.

Returns:

The run-output document as a dict.

hedbench.validation

Validate benchmark input and output documents against the JSON Schemas in json_schemas/.

The standardized format is not designed yet; see json_schemas/README.md for the current state and the planned document kinds.

hedbench.validation.validate_document(document, kind)

Validate one benchmark JSON document against its schema.

Parameters:

• document – The parsed JSON document (dict).

• kind – Which schema to validate against, e.g. “case”, “input”, “output”.

Returns:

A list of validation issues; empty if the document is valid.