python-dask-expr/python-dask-expr.changes

-------------------------------------------------------------------
Thu Dec  5 21:53:04 UTC 2024 - Ben Greiner <code@bnavigator.de>

- Update to 1.1.20
  * Fix value_counts with split_out != 1 (#1170) Patrick Hoefler
  * Remove recursion in task spec (#1158) Florian Jetter
  * Deprecated and remove from_legacy_dataframe usage (#1168)
    Patrick Hoefler
  * Remove from_dask_dataframe (#1167) Patrick Hoefler
  * Avoid exponentially growing graph for Assign-Projection
    combinations (#1164) Patrick Hoefler
  * Introduce more caching when walking the expression (#1165)
    Patrick Hoefler
  * Use Taskspec fuse implementation (#1162) Florian Jetter
  * Fix orphaned dependencies in Fused expression (#1163) Patrick
    Hoefler
- Add dask-expr-pr1173-blockwise.patch
  * Use new blockwise unpack collection in array
  * gh#dask/dask-expr#1173

-------------------------------------------------------------------
Fri Nov 22 12:06:57 UTC 2024 - Dirk Müller <dmueller@suse.com>

- reenable python313, numba is available

-------------------------------------------------------------------
Wed Nov 20 14:24:56 UTC 2024 - Benjamin Greiner <code@bnavigator.de>

- Disable python313: no numba, no dask

-------------------------------------------------------------------
Tue Nov 19 12:35:06 UTC 2024 - Dirk Müller <dmueller@suse.com>

- update to 1.1.19:
  * rebuild for dask 2024.11.2

-------------------------------------------------------------------
Tue Nov 12 08:47:46 UTC 2024 - Dirk Müller <dmueller@suse.com>

- update to 1.1.18:
  * Add support for Python 3.13
  * Migrate Blockwise to use taskspec
  * Migrate shuffle and merge to `P2PBarrierTask`
  * Improve Aggregation docstring explicitly mentionning
    SeriesGroupBy
  * Migrate P2P shuffle and merge to TaskSpec
  * Internal cleanup of P2P code
  * Fix meta calculation for to_datetime
  * Fix ``Merge`` divisions after filtering partitions

-------------------------------------------------------------------
Thu Oct 10 15:42:20 UTC 2024 - Dirk Müller <dmueller@suse.com>

- update to 1.1.15:
  * Add concatenate flag to .compute()
  * Import from tokenize

-------------------------------------------------------------------
Sun Sep  8 14:24:41 UTC 2024 - Ben Greiner <code@bnavigator.de>

- Update to 1.1.13
  * Use task-based rechunking as default (#1131) Hendrik Makait
  * Improve performance of DelayedsExpr through caching (#1132)
    Hendrik Makait
- Release 1.1.12
  * Fix concat axis 1 bug in divisions (#1128) Patrick Hoefler
  * Bump pyarrow>=14.0.1 minimum versions (#1127) James
    Bourbeau
  * Fix scalar detection of columns coming from sql (#1125)
    Patrick Hoefler

-------------------------------------------------------------------
Fri Aug 30 10:55:19 UTC 2024 - Ben Greiner <code@bnavigator.de>

- Skip some tests on 32-bit

-------------------------------------------------------------------
Thu Aug 29 14:38:23 UTC 2024 - Ben Greiner <code@bnavigator.de>

- Update to 1.1.11
  * Make split_out for categorical default smarter #1124 Patrick
    Hoefler
  * Avoid calling `array`` attribute on ``cudf.Series`` #1122
    `Richard (Rick) Zamora
  * Introduce ToBackend` expression #1115 `Richard (Rick) Zamora
  * Fix result index of merge #1121 Patrick Hoefler
  * Fix projection for Index class in read_parquet #1120 Patrick
    Hoefler
  * Register read_parquet` and `read_csv` as "dispatchable" #1114
    `Richard (Rick) Zamora
  * Fix merging when index name in meta missmatches actual name
    #1119 Patrick Hoefler
  * Fix tuples as on argument in merge #1117 Patrick Hoefler
  * Drop support for Python 3.9 #1109 Patrick Hoefler
- Release 1.1.10
  * Fixup remaining upstream failures #1111 Patrick Hoefler
  * Fix some things for pandas 3 #1110 Patrick Hoefler
- Release 1.1.9
  * Patch release for Dask 2024.7.0
- Release 1.1.8
  * Fix shuffle blowing up the task graph #1108 Patrick Hoefler
  * Link fix in readme #1107 Ben
  * Fix from_pandas with chunksize and empty df #1106 Patrick
    Hoefler
  * Fix deepcopying FromPandas class #1105 Patrick Hoefler
  * Skip test if optional xarray cannot be imported #1104 Sandro
- Release 1.1.7
  * Patch release for Dask 2024.7.0

-------------------------------------------------------------------
Mon Jul  8 04:34:30 UTC 2024 - Steve Kowalik <steven.kowalik@suse.com>

- Update to 1.1.6:
  * Patch release for Dask 2024.6.2
  * Fix resample divisions propagation
  * Fix categorize if columns are dropped
  * Fix projection to empty from_pandas
  * Fix meta for string accessors
  * Use is_categorical_dtype dispatch for sort_values
  * Fix read_csv with positional usecols
  * Fix isin for head computation
  * Fix isin with strings
  * Use ensure_deterministic kwarg instead of config
  * Add cache argument to lower_once
  * Fix non-integer divisions in FusedIO
  * Fix dropna before merge
  * Fix sort_values for unordered categories
  * Fix to_parquet in append mode
  * reduce pickle size of parquet fragments
  * Generalize get_dummies
  * Add support for DataFrame.melt
  * Fix default name conversion in ToFrame
  * Optimize when from-delayed is called
  * Fix delayed in fusing with multipled dependencies
  * Fix None min/max statistics and missing statistics generally
  * Fix xarray integration with scalar columns
  * Fix shape returning integer
  * Fix bug in Series reductions

-------------------------------------------------------------------
Mon Apr 29 12:43:13 UTC 2024 - Ben Greiner <code@bnavigator.de>

- Remove extra source file

-------------------------------------------------------------------
Fri Apr 26 18:00:13 UTC 2024 - Ben Greiner <code@bnavigator.de>

- Initial specfile for v1.0.13
- Required by and versions must by in sync with dask