Changes for version 0.49 - 2026-10-12
- Enhancements
- aggregate(): new method providing GROUP BY / measures across all backends. Pass by => [...], measures => [{col=>'amount',fn=>'sum',as=>'revenue'},...] plus any ordinary WHERE criteria. Functions: sum, avg, min, max, count, count_distinct. SQL pushdown for DBI backends (SQLite, CSV, DSN); Perl in-memory fold for slurp-only backends (XLSX, DBM::Deep, BerkeleyDB, JSON, HTML URL, Redis). HAVING, sort_by, limit, offset all supported. Empty-set normalisation: sum/count/count_distinct → 0; avg/min/max → undef. Cache, base_criteria, and fixate/no_fixate behave as for selectall_arrayref. Replaces three separate per-chart aggregation loops in Database::BI (the motivation: no shared definition of a measure was the single absence keeping the stack a file viewer rather than a BI tool).
- Query builder: group_by(), measure(), having() builder methods added. Terminal all()/first() delegate to aggregate() when measures are set. count() croaks when called after measure() with a clear error message.
- Tests
- t/aggregate.t: 70 tests covering SQL and Perl-fold paths, all measure functions, HAVING (including PostgreSQL-safe expression substitution), sort/limit/offset, empty-set and all-NULL groups, validation croaks, query-builder parity, CSV comment-row guard, DBM::Deep fold path, non-numeric warning on the Perl path.
Documentation
Modules
Read-only Database Abstraction Layer (ORM)
Fluent, chainable query builder for Database::Abstraction