Working with results¶
Stub
Final content will be a tour of the Result wrapper — how to get
your data out as pandas, Polars, Arrow, or parquet, and when to
pick which.
Topics planned:
- A cheat-sheet table of every
Resultmethod:arrow,df,to_pandas,to_polars,batches,head,column,write_parquet. - When the data lives in memory vs on disk (auto-spill,
inmem_row_limit). .batches()for streaming through results that don't fit in RAM.Result.write_parquet(path, layout="hats" | "single")— the two output shapes.- A "Want it faster?" callout on pandas vs Polars: lead with
pandas because that's what everyone knows, then nudge readers to
r.to_polars()for anything beyond inspection (filtering, group-by, joins on multi-million-row results are routinely 5–50× faster). Include a side-by-side pandas /astropy.Table/ Polars example for the same filter-group-mean operation.
For now, the Python API surface in the README lists every method.