New Proposal Seeks to Replace 'Archaic' SQL with Functional Paradigms

A new proposal for a modern relational query language argues that the enduring power of relational theory is often undermined by the clumsy and archaic implementation of SQL. The draft, published by the author of sporks.space, suggests that the gap between relational power and developer experience has contributed to the rise of NoSQL alternatives.

The author proposes a fundamental shift in how query languages are structured, advocating for a move away from PL/I-based syntax in favor of aesthetics found in modern languages like Rust, Python, C, or the ML family. Central to the proposal is the integration of functional programming paradigms, specifically the use of sum types, discriminated unions, and pattern matching. These tools would allow developers to handle mutually exclusive columns more effectively and significantly reduce the reliance on nullable fields, which often complicate data integrity.

The Case for Type Expressiveness

The push for these changes is framed against the backdrop of contemporary query language projects such as Acadia. The author's perspective is informed by extensive experience across major systems, including Oracle, SQL Server, SQLite, Db2, MySQL, and Postgres. Specifically, the author cites the limitations encountered in IBM i system administration functions as a primary catalyst for seeking better type expressiveness in relational contexts.

One specific technical recommendation is the implementation of 'discriminated union' foreign keys. This feature would enable a single many-to-many table to relate to multiple different tables simultaneously, eliminating the need to duplicate join tables for every unique relationship. Additionally, the author calls for more transparent query planners and improved user-defined types, or domains, noting that Postgres has historically remained closer to E.F. Codd's original relational vision than the standard SQL specification.

Industry Implications

This proposal highlights a growing tension in software engineering: the desire for the strict reliability of relational data paired with the flexibility of modern programming languages. By bridging the gap between declarative SQL and functional programming, the author suggests that the industry could drastically reduce boilerplate code and minimize common errors caused by null values. Such a shift would simplify complex schema designs and make the database layer feel like a first-class citizen in a modern development stack.

Future Outlook

While the draft serves as a wishlist for a theoretical modern language, it points toward a broader trend of rethinking data interaction. Whether these functional paradigms are adopted by existing giants like Postgres or drive the success of new projects like Acadia remains to be seen. For now, the proposal stands as a critique of the status quo and a blueprint for a more expressive, type-safe approach to relational data management.