pola-rs/polars py-1.8.0

Python Polars 1.8.0

on GitHub

✨ Enhancements

• Support arithmetic between Series with dtype list (#17823)
• Relaxed schema alignment for parquet file list read (#18803)
• Always preserve sorted flag for .dt.date (#18692)
• Enable additional ruff lint rule sets (#18721)
• Implement single inequality joins for join_where (#18727)

🐞 Bug fixes

• DataFrame plot was raising when some extra keywords were passed to encodings (e.g. x=alt.X(a, axis=alt.Axis(labelAngle=30))) (#18836)
• Respect strictness in list constructor (#18853)
• Properly broadcast array arithmetic (#18851)
• Throw error for comparison of unequal length series (#18816)
• Raise when parquet file has extra columns and no select() was done (#18843)
• Pass missing user params in write_csv (#18845)
• Improve join argument checks (#18847)
• Struct filter by index (#18778)
• Proper dtype casting for struct embedded categoricals in chunked categoricals (#18815)
• Fixed some error/assertion types (#18811)
• Remove panic in arr.to_struct (#18804)
• Allow empty sort by columns (#18774)
• Broadcast zip_with for structs (#18770)
• Dropped/shifted rows in parquet scan with streaming=True (#18766)
• Fix cum_max using exception text of cum_min for invalid dtype (#18780)
• Fix accidental raise on shape 1 (#18748)

📖 Documentation

• Fix link to issue tracker and code snippet format in GPU docs (#18850)
• Clarify documentation for schema in read_csv function (#18759)
• Fix literal type mapping example in lit docstrings (#18756)
• Refactor docs directory hierarchy (#18773)
• Minor improvements to contributing guide (#18777)
• Improve over docs, add example with order_by (#18796)
• Add documentation for beta gpu support (#18762)

🛠️ Other improvements

• Re-export PyO3 in polars-python crate (#18835)
• Make NodeTraverser struct public (#18822)
• Add panic to unchecked DataFrame constructors in debug mode (#18807)
• Fix parquet file metadata is dropped after first DSL->IR conversion (#18789)
• Remove extra hashmap construction in new-streaming parquet (#18792)
• Remove TODO comment regarding NumPy pinning (#18776)
• Remove unused methods (#18744)
• Make DataFrame a Vec of Column instead of Series (#18664)
• Run benchmark on PR labeled 'needs-bench' (#18737)
• Enable additional ruff lint rule sets (#18721)

Thank you to all our contributors for making this release possible!
@3ok, @Manishearth, @MarcoGorelli, @adamreeve, @alexander-beedie, @barak1412, @beckernick, @bradfordlynch, @coastalwhite, @deanm0000, @eitsupi, @i64, @itamarst, @mcrumiller, @nameexhaustion, @orlp, @r-brink, @ritchie46, @rodrigogiraoserrao, @squnit, @stinodego and @t-ded