Using JSON.parse(JSON.stringify()) is inefficient and can cause issues. This approach:

1. Has poor performance due to double serialization overhead
2. Will throw an error if result is undefined (when sql.execute() returns undefined for some queries)
3. Loses non-serializable properties like Buffers, Dates (converted to strings), and circular references
4. Is inconsistent with other database implementations (MySQL, PostgreSQL, SQL Server, ClickHouse) which all return the result directly

The reverted code with explicit structure and null coalescing (rows ?? [], metaData ?? []) was safer as it handled edge cases where the Oracle driver might return undefined or partial results.

This implementation is inconsistent with other database operations in the codebase. MySQL, PostgreSQL, SQL Server, and ClickHouse implementations all return the result directly without JSON serialization. Consider aligning the Oracle implementation with these patterns for consistency and maintainability.

The JSON.parse(JSON.stringify()) pattern introduces significant performance overhead, especially for large query results. This double serialization process is unnecessary and inefficient compared to either returning the result object directly or constructing a simple object with the needed properties.