PlanFrame documentation

PlanFrame is a typed, backend-agnostic planning layer for dataframe engines.

Choose a path:

• PlanFrame (core): build a statically-typed adapter for a backend engine.
• planframe-polars: use PlanFrame on top of Polars.
• planframe-pandas: use PlanFrame on top of Pandas.
• planframe-sparkless: use PlanFrame’s Spark-like interface on top of sparkless.

Install

Install an adapter package (recommended):

pip install planframe-polars
# or
pip install planframe-pandas
# or
pip install planframe-sparkless

Core-only (adapter authors / libraries):

pip install planframe

Start here

• CHANGELOG.md (authoritative per-release notes, including v1.3.0)
• Migrating since v1.1.0 (behavior and API changes from v1.2.0 through v1.3.0)
• PlanFrame (core)
• planframe-polars
• planframe-pandas
• planframe-sparkless
• PySpark-like API (planframe.spark)
• Pandas-like API (planframe.pandas)
• Adapter capability matrix
• Stability & compatibility

What’s new (high-level)

For v1.3.0 and earlier releases, read CHANGELOG.md. Upgrade how-tos and versioned API notes live in Migrating since v1.1.0.

Adding a new adapter

If you’re creating a new adapter, start with:

• Creating an adapter
• Core layout (how Frame, compilation, and execute_plan are organized in the codebase)
• Adapter docs template