Recent pandas updates
Section
Features
- Add inline examples for Multiindexing/Levels section (#64867) (759a5fd)
- enable python3.15 job (#66473) (7bb284a)
Section
Fixes
- BUG: DataFrame.stack raising a bare AssertionError on duplicate levels (GH#66588) (#66590) (9e01ba9)
- Fix halflife math rendering in DataFrame.ewm (#66381) (300a0cd)
- BUG: Fix merge not upcasting datetime64 join key to highest resolution for inner/left joins (#65102) (4ced8bc)
- BUG: read_csv parallel default ignored CPU affinity and cgroup limits (#66613) (310b256)
- Fix performance regression in string construction from GH-65831 (#66351) (8a8188e)
Section
Improvements
- DEPR: convert_dtypes boolean keywords (#65169) (71959b8)
- consolidate read_csv token metadata into a single word_ends array (#66276) (5ce9fa7)
- read_csv integer fast path using fast_float's digit helpers (#66487) (2454e77)
- single-pass composite-key factorization for multi-key merges (#66124) (34976fd)
- merge_asof with by (#66121) (f0ed68a)
- read_csv string columns, cutting per-column-chunk GIL work (#66619) (1d6b36e)
Section
Breaking Changes
- remove pyarrow 24.0 restriction for typing (#66563) (340b18d)