SQL is syntactic sugar for relational algebra

Original Article ↗ Hacker News Discussion ↗

Overall view of SQL

  • Widely used and powerful, but many see the syntax as messy, inconsistent, and “backwards,” especially the SELECT-before-FROM ordering and odd scoping rules.
  • Others find it reasonably learnable, expressive, and preferable to JSON-based or proprietary query syntaxes; for many, problems stem more from poor schema design and misuse than from SQL itself.
  • Comparison to C: good to know deeply, but some prefer “managed” abstractions; critics counter that ORMs are leaky and often underpowered.

Syntax, ergonomics, and tools

  • Complaints: non-intuitive parse order, difficulty reading complex nested queries, hard autocomplete because sources come after projections, and dangerous commands like DELETE FROM without guardrails.
  • Mitigations: use of CTEs, table-valued functions, temp tables, transactions, and client safeguards (WHERE-required deletes, special flags).
  • Layout, comments, and ER diagrams are seen as crucial for maintaining complex SQL.

Relational algebra, theory, and the standard

  • Many agree SQL is not clean syntactic sugar over relational algebra: bags vs sets, NULL semantics, ordering, and many “non-relational doodads.”
  • References to Codd’s relational model, predicate calculus, Datalog, MonetDB’s IR, and the “Third Manifesto” highlight a desire for languages closer to formal relational theory.
  • The SQL standard is viewed by some as overly complex and effectively a moat that raises the barrier for new database implementations.

Alternatives and related languages

  • Mentioned alternatives or complements: PRQL, XTQL, LINQ-style comprehensions, Malloy, CozoDB, dataframe libraries (Pandas, Polars, Ibis), and D-style languages (though no popular implementation).
  • These often improve composability, left-to-right reading, and integration with host languages, but typically lack full DML support, vendor-specific features, or maturity.

Learning, pedagogy, and practice

  • Several recount taking years to “really get” SQL, often triggered by performance-heavy or large-scale work.
  • Formal DB education is uneven; some criticize highly theoretical teaching (including some relational-theory texts), others praise accessible introductions and practical design books.
  • Strong theme: understanding relational theory and good schema design drastically improves SQL, but SQL’s warts still limit what well-designed RDBMS engines could expose.