TableAs

TableAs is a standard and library for representing relational data as first-class objects. It provides a uniform API for tabular data, with typed columns, row-level operations, and metadata about schemas and relationships. The design emphasizes immutability and lazy evaluation to enable efficient processing of large datasets and reproducible analyses.

The project originated at DataForge Labs in the early 2010s as an attempt to unify CSV-like interchange with database-style semantics. The format and accompanying library matured through community contributions and reached a stable 1.0 release in 2018, with continued updates focused on interoperability and streaming support.

Core features include explicit column typings, schema validation, support for relational operations such as joins and group-by aggregations, time-aware and hierarchical data types, streaming and incremental computation, and compatibility adapters for CSV, JSON, Parquet, and database connections. The design prioritizes readability and portability, enabling dashboards, notebooks, and data pipelines to share a common table abstraction.

Typical usage involves creating a TableAs instance from a data source, applying filtering and projection, performing aggregations, and combining results via joins. Because of lazy evaluation, the full computation is realized only when needed. The format also supports metadata about provenance and lineage, aiding reproducibility.

See also: CSV, JSON, Parquet, relational algebra, and various DataFrame libraries used in data science.