|
STAR
Cal STAR — firmware, DAQ server, and engine design
|
Provenance is mandatory, and survives a round trip. Phase 02's headline exit criterion is that *a parameter with no source is a validation error rather than a default*. That is what these tests hold. The rest check that the record survives serialisation intact, because provenance that is lost on the first save is provenance that does not exist.
| None test_model_param.test_a_parameter_with_no_source_is_rejected | ( | ) |
The criterion, stated directly. A missing source must not default to anything -- not to ESTIMATED, not to DEFAULT. Defaulting it would mean every un-sourced number silently acquired a plausible-looking provenance, which is worse than having none at all.
Definition at line 25 of file test_model_param.py.
| None test_model_param.test_a_report_can_say_what_a_result_rests_on | ( | ) |
The point of carrying provenance at all. "Nine measured, four datasheet, two guesses" is a different claim from a bare number, and the difference should never be invisible in a design review.
Definition at line 130 of file test_model_param.py.
| None test_model_param.test_a_source_must_be_one_of_the_declared_kinds | ( | ) |
Definition at line 36 of file test_model_param.py.
| None test_model_param.test_an_unknown_unit_is_caught_at_construction | ( | ) |
Definition at line 49 of file test_model_param.py.
| None test_model_param.test_assumed_flags_the_values_nobody_established | ( | ) |
Definition at line 123 of file test_model_param.py.
| None test_model_param.test_optional_fields_are_omitted_rather_than_nulled | ( | ) |
A terse parameter stays terse when written back.
Definition at line 81 of file test_model_param.py.
| None test_model_param.test_provenance_is_ranked_so_measurements_win | ( | ) |
Definition at line 117 of file test_model_param.py.
| None test_model_param.test_round_trip_through_json_is_exact | ( | ) |
A parameter survives the full save/load path unchanged.
Definition at line 68 of file test_model_param.py.
| None test_model_param.test_si_bounds_are_in_canonical_units | ( | ) |
Definition at line 109 of file test_model_param.py.
| None test_model_param.test_summary_always_reports_every_kind | ( | ) |
A stable table, so a report does not gain and lose columns.
Definition at line 152 of file test_model_param.py.
| None test_model_param.test_the_source_must_be_the_enum_not_a_string | ( | ) |
Constructed directly, a stringly-typed source is a type error.
Definition at line 43 of file test_model_param.py.
| None test_model_param.test_uncertainty_bounds | ( | ) |
Definition at line 87 of file test_model_param.py.
| None test_model_param.test_uncertainty_cannot_be_negative | ( | ) |
Definition at line 104 of file test_model_param.py.
| None test_model_param.test_uncertainty_kind_is_checked | ( | str | kind | ) |
Definition at line 99 of file test_model_param.py.
| None test_model_param.test_value_is_stored_as_authored_and_converted_on_demand | ( | ) |
Round-trip fidelity depends on keeping the authored form. Converting on construction would mean a config written in psi came back as a float with a trail of decimals -- a diff on every save, and a file that no longer looks like what someone wrote.
Definition at line 54 of file test_model_param.py.