Problem
Cases with kind: narrative have no operation and no expected: list. The harness counts them as passing:
But 5 of those 44 are narrative-only (documentation):
- name: edm_namespace_is_reserved
kind: narrative
desc: "The Edm namespace is reserved for OData primitive and abstract types..."
- name: date_types_use_proleptic_gregorian
kind: narrative
desc: "Edm.Date and Edm.DateTimeOffset use the proleptic Gregorian calendar..."
Their traces are empty (no traces) — there's nothing to validate.
Impact
"44 cases passed" suggests 44 behavioral validations ran. In reality only 39 actually tested anything. This misrepresents coverage.
Suggestion
Report separately:
39 cases passed, 5 narrative (no assertions)
Or add a CaseStatus::Narrative / CaseStatus::Skipped variant to distinguish them from actual test executions.
Problem
Cases with
kind: narrativehave no operation and noexpected:list. The harness counts them as passing:But 5 of those 44 are narrative-only (documentation):
Their traces are empty
(no traces)— there's nothing to validate.Impact
"44 cases passed" suggests 44 behavioral validations ran. In reality only 39 actually tested anything. This misrepresents coverage.
Suggestion
Report separately:
Or add a
CaseStatus::Narrative/CaseStatus::Skippedvariant to distinguish them from actual test executions.