Skip to content

Latest commit

 

History

History
228 lines (163 loc) · 9.4 KB

changes.md

File metadata and controls

228 lines (163 loc) · 9.4 KB

Dask-expr

v1.1.21

  • Fixup install paths (:pr:1179) Patrick Hoefler_
  • Remove custom read-csv stuff (:pr:1178) Patrick Hoefler_
  • Fix assign optimization when overwriting columns (:pr:1176) Patrick Hoefler_
  • Propagate group_keys in DataFrameGroupBy (:pr:1174) Tom Augspurger_
  • Use new blockwise unpack collection in array (:pr:1173) James Bourbeau_

v1.1.20

  • Fix value_counts with split_out != 1 (:pr:1170) Patrick Hoefler_
  • Remove recursion in task spec (:pr:1158) Florian Jetter_
  • Deprecated and remove from_legacy_dataframe usage (:pr:1168) Patrick Hoefler_
  • Remove from_dask_dataframe (:pr:1167) Patrick Hoefler_
  • Avoid exponentially growing graph for Assign-Projection combinations (:pr:1164) Patrick Hoefler_
  • Introduce more caching when walking the expression (:pr:1165) Patrick Hoefler_
  • Use Taskspec fuse implementation (:pr:1162) Florian Jetter_
  • Fix orphaned dependencies in Fused expression (:pr:1163) Patrick Hoefler_

v1.1.19

v1.1.18

v1.1.17

  • Add support for Python 3.13 (:pr:1160) James Bourbeau_
  • Migrate Blockwise to use taskspec (:pr:1159) Florian Jetter_
  • Migrate shuffle and merge to P2PBarrierTask (:pr:1157) Hendrik Makait_
  • Improve Aggregation docstring explicitly mentionning SeriesGroupBy (:pr:1156) Guillaume Eynard-Bontemps_
  • Migrate P2P shuffle and merge to TaskSpec (:pr:1155) Hendrik Makait_
  • Internal cleanup of P2P code (:pr:1154) Hendrik Makait_
  • Fix meta calculation for to_datetime (:pr:1153) Patrick Hoefler_
  • Fix Merge divisions after filtering partitions (:pr:1152) Richard (Rick) Zamora_

v1.1.16

v1.1.15

  • Add concatenate flag to .compute() (:pr:1138) Hendrik Makait_

v1.1.14

  • Import from tokenize (:pr:1133) Patrick Hoefler_

v1.1.14

  • Import from tokenize (:pr:1133) Patrick Hoefler_

v1.1.13

v1.1.12

  • Fix concat axis 1 bug in divisions (:pr:1128) Patrick Hoefler_
  • Bump pyarrow>=14.0.1 minimum versions (:pr:1127) James Bourbeau_
  • Fix scalar detection of columns coming from sql (:pr:1125) Patrick Hoefler_

v1.1.11

  • Make split_out for categorical default smarter (:pr:1124) Patrick Hoefler_
  • Avoid calling array attribute on cudf.Series (:pr:1122) Richard (Rick) Zamora_
  • Introduce ToBackend expression (:pr:1115) Richard (Rick) Zamora_
  • Fix result index of merge (:pr:1121) Patrick Hoefler_
  • Fix projection for Index class in read_parquet (:pr:1120) Patrick Hoefler_
  • Register read_parquet and read_csv as "dispatchable" (:pr:1114) Richard (Rick) Zamora_
  • Fix merging when index name in meta missmatches actual name (:pr:1119) Patrick Hoefler_
  • Fix tuples as on argument in merge (:pr:1117) Patrick Hoefler_
  • Drop support for Python 3.9 (:pr:1109) Patrick Hoefler_

v1.1.10

  • Fixup remaining upstream failures (:pr:1111) Patrick Hoefler_
  • Fix some things for pandas 3 (:pr:1110) Patrick Hoefler_

v1.1.9

  • Patch release for Dask 2024.7.0

v1.1.8

  • Fix shuffle blowing up the task graph (:pr:1108) Patrick Hoefler_
  • Link fix in readme (:pr:1107) Ben_
  • Fix from_pandas with chunksize and empty df (:pr:1106) Patrick Hoefler_
  • Fix deepcopying FromPandas class (:pr:1105) Patrick Hoefler_
  • Skip test if optional xarray cannot be imported (:pr:1104) Sandro_

v1.1.7

  • Patch release for Dask 2024.7.0

v1.1.6

v1.1.5

  • Patch release for Dask 2024.6.2

v1.1.4

v1.1.3

  • Fix resample divisions propagation (:pr:1075) Patrick Hoefler_
  • Fix categorize if columns are dropped (:pr:1074) Patrick Hoefler_

v1.1.2

  • Fix projection to empty from_pandas (:pr:1072) Patrick Hoefler_
  • Fix meta for string accessors (:pr:1071) Patrick Hoefler_
  • Use is_categorical_dtype dispatch for sort_values (:pr:1070) Richard (Rick) Zamora_

v1.1.1

  • Fix read_csv with positional usecols (:pr:1069) Patrick Hoefler_
  • Fix isin for head computation (:pr:1068) Patrick Hoefler_
  • Fix isin with strings (:pr:1067) Patrick Hoefler_
  • Use ensure_deterministic kwarg instead of config (:pr:1064) Florian Jetter_
  • Add cache argument to lower_once (:pr:1059) Richard (Rick) Zamora_
  • Fix non-integer divisions in FusedIO (:pr:1063) Patrick Hoefler_
  • Fix dropna before merge (:pr:1062) Patrick Hoefler_
  • Fix sort_values for unordered categories (:pr:1058) Patrick Hoefler_
  • Fix to_parquet in append mode (:pr:1057) Patrick Hoefler_

v1.1.0

  • Add a bunch of docs (:pr:1051) Patrick Hoefler_
  • reduce pickle size of parquet fragments (:pr:1050) Florian Jetter_
  • Generalize get_dummies (:pr:1053) Richard (Rick) Zamora_
  • Fixup failing test (:pr:1052) Patrick Hoefler_
  • Add support for DataFrame.melt (:pr:1049) Richard (Rick) Zamora_
  • Fix default name conversion in ToFrame (:pr:1044) Richard (Rick) Zamora_
  • Optimize when from-delayed is called (:pr:1048) Patrick Hoefler_

v1.0.14

  • Fix delayed in fusing with multipled dependencies (:pr:1038) Patrick Hoefler_
  • Fix drop with set (:pr:1047) Patrick Hoefler_
  • Fix None min/max statistics and missing statistics generally (:pr:1045) Patrick Hoefler_
  • Fix xarray integration with scalar columns (:pr:1046) Patrick Hoefler_
  • Fix shape returning integer (:pr:1043) Patrick Hoefler_
  • Fix bug in Series reductions (:pr:1041) Richard (Rick) Zamora_

v1.0.13

  • Fix shuffle after set_index from 1 partition df (:pr:1040) Patrick Hoefler_
  • Fix loc slicing with Datetime Index (:pr:1039) Patrick Hoefler_
  • Fix loc accessing index for element wise op (:pr:1037) Patrick Hoefler_
  • Fix backend dispatching for read_csv (:pr:1028) Richard (Rick) Zamora_
  • Add cudf support to to_datetime and _maybe_from_pandas (:pr:1035) Richard (Rick) Zamora_

v1.0.12

  • Move IO docstrings over (:pr:1033) Patrick Hoefler_
  • Fuse more aggressively if parquet files are tiny (:pr:1029) Patrick Hoefler_
  • Add nr of columns to explain output for projection (:pr:1030) Patrick Hoefler_
  • Fix error in analyze for scalar (:pr:1027) Patrick Hoefler_
  • Fix doc build error (:pr:1026) Patrick Hoefler_
  • Add docs for usefule optimizer methods (:pr:1025) Patrick Hoefler_
  • Rename uniuqe_partition_mapping property and add docs (:pr:1022) Patrick Hoefler_
  • Fix read_parquet if directory is empty (:pr:1023) Patrick Hoefler_
  • Fix assign after set index incorrect projections (:pr:1020) Patrick Hoefler_
  • Use implicit knowledge about divisions for efficient grouping (:pr:946) Florian Jetter_
  • Simplify dtype casting logic for shuffle (:pr:1012) Patrick Hoefler_
  • Fix column projections in merge when suffixes are relevant (:pr:1019) Patrick Hoefler_

v1.0.11

  • Fix unique with numeric columns (:pr:1017) Patrick Hoefler_
  • Fix projection for rename if projection isn't renamed (:pr:1016) Patrick Hoefler_
  • Fix head for npartitions=-1 and optimizer step (:pr:1014) Patrick Hoefler_
  • Deprecate to/from_dask_dataframe API (:pr:1001) Richard (Rick) Zamora_

v1.0.10

  • Make setattr work (:pr:1011) Patrick Hoefler_
  • Adjust version number in changes Patrick Hoefler_

v1.0.9

  • Add support for named aggregations in groupby(...).aggregate() (:pr:1009) Patrick Hoefler_

v1.0.7

  • Fix meta calculation in drop_duplicates to preserve dtypes (:pr:1007) Patrick Hoefler_

v1.0.6

  • Fix pyarrow fs reads for list of directories (:pr:1006) Patrick Hoefler_
  • Register json and orc APIs for "pandas" dispatch (:pr:1004) Richard (Rick) Zamora_
  • Rename overloaded to/from_dask_dataframe API (:pr:987) Richard (Rick) Zamora_
  • Fix zero division error when reading index from parquet (:pr:1000) Patrick Hoefler_
  • Start building and publishing conda nightlies (:pr:986) Charles Blackmon-Luca_
  • Set divisions with divisions already known (:pr:997) Florian Jetter_
  • Nicer read_parquet prefix (:pr:998) Florian Jetter_
  • Reduce coverage target a little bit (:pr:999) Patrick Hoefler_

v1.0.5

  • Ensure that repr doesn't raise if an operand is a pandas object (:pr:996) Florian Jetter_
  • Allow passing of boolean index for column index in loc (:pr:995) Florian Jetter_
  • Update pyproject.toml (:pr:994) Florian Jetter_
  • Fix SettingWithCopyWarning in merge.py (:pr:990) Miles
  • Ensure drop matches column names exactly (:pr:992) Florian Jetter_
  • Support prefix argument in from_delayed (:pr:991) Richard (Rick) Zamora_
  • Visual ANALYZE (:pr:889) Hendrik Makait_

v1.0.4

  • Ensure wrapping an array when comparing to Series works if columns are empty (:pr:984) Florian Jetter_
  • Remove keys() (:pr:983) Patrick Hoefler_
  • Fix some reset_index optimization issues (:pr:982) Patrick Hoefler_
  • Fix concat of series objects with column projection (:pr:981) Patrick Hoefler_
  • Raise better error for repartition on divisions with unknown divisions (:pr:980) Patrick Hoefler_

v1.0.3

  • Support for dask==2023.3.1

v1.0.2

  • Revert enabling pandas cow (:pr:974) Florian Jetter_
  • Fixup predicate pushdown for query 19 (:pr:973) Patrick Hoefler_
  • Fixup set_index with one partition but more divisions by user (:pr:972) Patrick Hoefler_
  • Implement custom reductions (:pr:970) Patrick Hoefler_
  • Fix unique with shuffle and strings (:pr:971) Patrick Hoefler_
  • Fixup filter pushdown through merges with ands and column reuse (:pr:969) Patrick Hoefler_

v1.0.0

Initial stable release