Local migration rig Pandas → Polars
Run both transformations on the same redacted fixture. Get one go/no-go report for values, schema, row order, time, memory, and streaming-plan risk.
Seeded run / 003 faults
Choose a circuit. The live, keyboard-operable report shows the evidence the CLI returns. Measured failures and plan heuristics never share a label.
Values differ at row 1, column 2.
No plan markers detected. This does not prove every operator streamed.
One command / evidence first
Point a small TOML file at your Pandas and Polars functions plus representative CSV or Parquet fixtures.
Each engine gets an isolated Python process. Fixture size is enforced before either import or transformation runs.
Review measured parity and benchmark samples separately from version-sensitive explain-plan heuristics.
Rust binary / Python-aware
The binary embeds its Python bridge. Your data and transformations stay where they are.
cargo install --git https://github.com/B-Divyesh/sf-data-engine-switchboardswitchboard init && switchboard assessField kit / one-time
The core assessor, JSON export, and every safety check are MIT-licensed. A $29 one-time Field Kit unlocks CI policy presets, a representative-fixture matrix, rollout checkpoints, and a printable migration review.
Sociobot/Dodo is the merchant of record. Refunds are handled there and revoke the license automatically.