Roadmap¶

Current release: 1.19.2. PydanTable is stable 1.x under semantic versioning.

This page lists forward-looking work. Shipped history and phase checklists live in Roadmap history.

Product direction¶

The public API stays SQLModel-like:

DataFrameModel as the primary service-oriented table type
Typed Expr transforms with schema propagation
Default Rust + Polars execution; optional SQL, Mongo, and Spark engines
Explicit materialization (rows, column dicts, Arrow/Polars when extras installed)

Area	Direction
Engine parity	Keep SQL/Mongo/Spark capability matrix and error messages aligned (Engine parity)
Docs & DX	Continue adoption-focused docs, API reference coverage, and verified examples
Polars alignment	Close high-value Expr/table gaps where schema story stays clear (Polars alignment)
FastAPI	Optional ecosystem helpers per demand (Enhancements)

Non-string map keys (dict[int, T], heterogeneous Arrow maps) — deferred; see Supported types
JSON-native incremental async iterators without full dict materialization first
Map transforms (map_filter, map_entries_sorted, …) where Polars ops are stable
Window extras: percent_rank, ntile, cume_dist, first_value/last_value with ignore-nulls
Table analytics helpers: multi-column quantile/median, corr/cov (with materialization cost documented)

Full candidate list: Future method candidates in history.

Not committed to a release train:

Distributed Spark execution — compile plans to JVM Spark (separate from today's PySpark façade)
SQL-backed plan execution — lower logical plans to database SQL beyond current lazy-SQL bridge scope

Breaking removals planned for 2.0.0 (no fixed date): see Migration guide and Versioning.

Open a GitHub issue with your use case, or contribute via CONTRIBUTING.md.