What question did this study set out to answer?

The aim is to develop a SQL equivalence checker that accounts for floating-point rounding and runtime errors.

August 11, 2026Open Access

Precision- and Error-Aware SQL Equivalence: Deciding Rewrites Under IEEE-754 Rounding and Runtime Errors

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial demonstrates SQL equivalence under floating-point and error semantics, highlighting issues in production optimizers.

The aim is to develop a SQL equivalence checker that accounts for floating-point rounding and runtime errors.
Developed a precision- and error-aware SQL equivalence checker using SMT theory of floating point.
Tested the checker against a production optimizer, auditing 391 pairs of assertions from Calcite's checkSimplify tests.
Validated examples of reassociation, distribution, and cancellation with Float32 counterexamples.
Disproved certain rewrites under precision and error semantics with concrete counterexamples.
The checker correctly flagged mistakes in Apache Calcite's RexSimplify, proving its utility in bug-finding.
Agreed with all 26 assertions from the Calcite test suite without contradictions, highlighting gaps in existing tests.

Nachiket Lele (2026) studied this question.

View Full Paper

Ask AI

Bookmark