This release introduces support for Polars API, a new query compiler for small data,
more functions that can use dynamic partitioning, as well as several bug fixes.
Key Features and Updates Since 0.31.0
- Stability and Bugfixes
- FIX-#0000: Fix type hint (#7343)
- FIX-#7113: Fix docstring overrides for subclasses (#7354)
- FIX-#7134: Use a separate docstring class for
BasePandasDataset
(#7353) - FIX-#7329: Do not sort columns on
df.update
(#7330) - FIX-#7351: Add ipython method calls to non-lookup list (#7352)
- FIX-#7355: Cpu count would be set incorrectly on a cluster (#7356)
- FIX-#7357: Fix
NoAttributeError
onDataFrame.copy
(#7358) - FIX-#7371: Fix inserting datelike values into a DataFrame (#7372)
- FIX-#7373: Try a previous version of
motoserver/moto
service, pin to 5.0.13 (#7374) - FIX-#7379: Fix
__imul__
performing addition instead of multiplication (#7380) - FIX-#7387: Limit the number of pytest workers for tests with Ray engine on Windows (#7388)
- FIX-#7389: Fix uploading artifacts (#7390)
- Refactor Codebase
- REFACTOR-#0000: Update copyright date (#7333)
- Documentation improvements
- New Features
- FEAT-#4605: Add native query compiler (#7259)
- FEAT-#7308: Interoperability between query compilers (#7376)
- FEAT-#7331: Initial Polars API (#7332)
- FEAT-#7337: Using dynamic partitionning in
broadcast_apply
(#7338) - FEAT-#7340: Add more granular lazy flags to query compiler (#7348)
- FEAT-#7368: Add a new environment variable for using dynamic partitioning (#7369)
Contributors
@MortalHappiness
@Retribution98
@YarShev
@ZhipengXue97
@anmyachev
@arunjose696
@devin-petersohn
@likawind
@sfc-gh-joshi
@sfc-gh-mvashishtha