Jaunty Known Limitations

Last reviewed: 2026-03-03

This file tracks currently known product limitations that impact API completeness, consistency, or production adoption.

High Priority

1. Source-generated ordinal caching is not schema-shape aware

• Risk: incorrect column mapping when the same entity type is read with different column order/shape across executions.
• Scope: generated IMapped<T> path.
• Status: active fix in P0 roadmap.

2. Async API requires DbConnection in most methods

• Sync API is IDbConnection-based, async API is mostly DbConnection-based.
• Impact: inconsistent ergonomics for consumers using interface-only abstractions.

3. Reflection still exists in core assembly paths

• Runtime mapper/binder resolution and extension bootstrap still use reflection in Jaunty core.
• Impact: reduces confidence in strict NativeAOT/zero-reflection claims.

Medium Priority

4. Fluent API parity gaps

• IJoinedQuery3<T1,T2,T3> is missing async/paging/sorting parity with 2-way join flows.
• Some distinct/CTE method families are still inconsistent.

5. Scaffolding lacks navigation property generation

• Foreign key metadata is discovered but not fully emitted as navigation properties.

6. Benchmark comparability quality is uneven

• Some benchmark result sets include incomplete competitor rows.
• Impact: weakens external performance claims unless clearly scoped.

Low Priority / Design Tradeoffs

7. Provider-specific scalar type differences

• COUNT(*) type differs by provider (int vs long).
• Workaround: use long for cross-provider scalar counts.

8. IMapped<T>.ReadEntity is strict-shape oriented

• Custom mapper assumes required columns exist.
• Workaround: use QueryPartial* with projection-safe mapper for partial shapes.

Tracking

• Active roadmap: the production-readiness tasklist, held in the private work repository
• Issues: https://github.com/extrode/jaunty/issues