Rhyme: A Multi-Paradigm Declarative Query Language
summary
Rhyme is an implemented declarative language for nested data that expresses aggregations, group-bys, joins and tensor ops in a single JSON-like syntax and compiles to JS and SQL codegen. It uses named generators for co-iteration/inner-join semantics, a deep path variables generator for shape-polymorphic queries, and maps tensor expressions to einsum-like notation. A prototype generates an IR and produces SQL that matches state-of-the-art compiled DB engine performance in preliminary experiments and is available open source.
AI-written summary · about
chapters (10)
- 0:00Rye introduction
- 1:44Core operations and nested queries
- 3:32Generators and aggregation
- 5:58Named generators and iterations
- 8:04Tensor operations and Einstein notation
- 10:10EinOps compatibility and examples
- 11:54Deep path and polymorphic queries
- 12:16Execution model and IR
- 13:39Performance results and SQL
- 14:00Conclusion and open-source
related talks
48mCompositional, Expressive, Performant Database Queries: Choose 3
32mArel: The Ruby Relational Algebra Library
20mFunctional collection programming with semi-ring dictionaries
37mFrom NoSQL to Mo' SQL - Adding Structure and Queriability to Riak
05mRelational Programming
43mDesigning Cypher - a Graph Query Language
52mData Over Code: A Short Hyperbole
37mRhino: low-latency key-value database in Rust