Description
The LMDB theme benchmark only checked the number of result rows. Scalar aggregate queries such as SELECT (COUNT(...) AS ?count) return one row even when the count binding is wrong, so regressions in the returned count value can go unnoticed.
Desired behavior
Track expected scalar ?count values for the theme query catalog and have ThemeQueryBenchmark.executeQuery() verify the literal count binding for queries that project AS ?count, while preserving the existing result-row count checks.
Acceptance criteria
- Catalog records expected scalar count binding values for every theme/query index where applicable.
- LMDB benchmark fails descriptively if a scalar-count query returns zero or multiple rows.
- LMDB benchmark fails descriptively if
?count is missing, non-literal, or differs from the catalog value.
- Focused smoke coverage exercises a scalar count query, for example PHARMA q0 expected
?count = 18.
Notes
This catches regressions where the benchmark still observes one result row but the actual aggregate count is wrong.
Description
The LMDB theme benchmark only checked the number of result rows. Scalar aggregate queries such as
SELECT (COUNT(...) AS ?count)return one row even when the count binding is wrong, so regressions in the returned count value can go unnoticed.Desired behavior
Track expected scalar
?countvalues for the theme query catalog and haveThemeQueryBenchmark.executeQuery()verify the literal count binding for queries that projectAS ?count, while preserving the existing result-row count checks.Acceptance criteria
?countis missing, non-literal, or differs from the catalog value.?count = 18.Notes
This catches regressions where the benchmark still observes one result row but the actual aggregate count is wrong.