Virtual Materialized Views for Atlassian Forge SQL: A Two-Level Caching Approach in forge-sql-orm

Atlassian Forge SQL (backed by TiDB) lacks native materialized view support, making expensive join/aggregation queries prone to the platform's 5-second timeout and 16MB memory limits. The forge-sql-orm library addresses this with a two-level caching approach—an invocation-scoped in-memory cache and a persistent cross-invocation cache via @forge/kvs—that acts as virtual materialized views at the application layer. Cache-aware write operations and automatic invalidation keep read-heavy features responsive without manual synchronization logic.