Changelog

2021.07.1

Released on July 23, 2021

Note

This is the last release with support for NumPy 1.17 and pandas 0.25. Beginning with the next release, NumPy 1.18 and pandas 1.0 will be the minimum supported versions.

• Make array assert_eq check dtype (#7903`) Julia Signell

• Remove skips from doctests (6 of 6) (GH#7863) Zhengnan Zhao

• Remove experimental feature warning from actors docs (GH#7925) Matthew Rocklin

• Remove skips from doctests (2 of 6) (GH#7873) Zhengnan Zhao

• Separate out Array and Bag API (GH#7917) Julia Signell

• Implement lazy Array.__iter__ (GH#7905) Julia Signell

• Clean up places where we inadvertently iterate over arrays (GH#7913) Julia Signell

• Add numeric_only kwarg to DataFrame reductions (GH#7831) Julia Signell

• Add pytest marker for GPU tests (GH#7876) Charles Blackmon-Luca

• Add support for histogram2d in dask.array (GH#7827) Doug Davis

• Remove skips from doctests (1 of 6) (GH#7874) Zhengnan Zhao

• Add node size scaling to the Graphviz output for the high level graphs (GH#7869) Freyam Mehta

• Update old Bokeh links (GH#7915) Bryan Van de Ven

• Temporarily pin fastparquet in CI (GH#7907) James Bourbeau

• Add dask.array import to progress bar docs (GH#7910) Fabian Gebhart

• Use separate files for each DataFrame API function and method (GH#7890) Julia Signell

• Fix pyarrow-dataset ordering bug (GH#7902) Richard (Rick) Zamora

• Generalize unique aggregate (GH#7892) GALI PREM SAGAR

• Raise NotImplementedError when using pd.Grouper (GH#7857) Ruben van de Geer

• Add aggregate_files argument to enable multi-file partitions in read_parquet (GH#7557) Richard (Rick) Zamora

• Un-xfail test_daily_stock (GH#7895) James Bourbeau

• Update access configuration docs (GH#7837) Naty Clementi

• Use packaging for version comparisons (GH#7820) Elliott Sales de Andrade

• Handle infinite loops in merge_asof (GH#7842) gerrymanoim

2021.07.0

Released on July 9, 2021

• Include fastparquet in upstream CI build (GH#7884) James Bourbeau

• Blockwise: handle non-string constant dependencies (GH#7849) Mads R. B. Kristensen

• fastparquet now supports new time types, including ns precision (GH#7880) Martin Durant

• Avoid ParquetDataset API when appending in ArrowDatasetEngine (GH#7544) Richard (Rick) Zamora

• Add retry logic to test_shuffle_priority (GH#7879) Richard (Rick) Zamora

• Use strict channel priority in CI (GH#7878) James Bourbeau

• Support nested dask.distributed imports (GH#7866) Matthew Rocklin

• Should check module name only, not the entire directory filepath (GH#7856) Genevieve Buckley

• Updates due to https://github.com/dask/fastparquet/pull/623 (GH#7875) Martin Durant

• da.eye fix for chunks=-1 (GH#7854) Naty Clementi

• Temporarily xfail test_daily_stock (GH#7858) James Bourbeau

• Set priority annotations in SimpleShuffleLayer (GH#7846) Richard (Rick) Zamora

• Blockwise: stringify constant key inputs (GH#7838) Mads R. B. Kristensen

• Allow mixing dask and numpy arrays in @guvectorize (GH#6863) Julia Signell

• Don’t sample dict result of a shuffle group when calculating its size (GH#7834) Florian Jetter

• Fix scipy tests (GH#7841) Julia Signell

• Deterministically tokenize datetime.date (GH#7836) James Bourbeau

• Add sample_rows to read_csv-like (GH#7825) Martin Durant

• Fix typo in config.deserialize docstring (GH#7830) Geoffrey Lentner

• Remove warning filter in test_dataframe_picklable (GH#7822) James Bourbeau

• Improvements to histogramdd (for handling inputs that are sequences-of-arrays). (GH#7634) Doug Davis

• Make PY_VERSION private (GH#7824) James Bourbeau

2021.06.2

Released on June 22, 2021

• layers.py compare parts_out with set(self.parts_out) (GH#7787) Genevieve Buckley

• Make check_meta understand pandas dtypes better (GH#7813) Julia Signell

• Remove “Educational Resources” doc page (GH#7818) James Bourbeau

2021.06.1

Released on June 18, 2021

• Replace funding page with ‘Supported By’ section on dask.org (GH#7817) James Bourbeau

• Add initial deprecation utilities (GH#7810) James Bourbeau

• Enforce dtype conservation in ufuncs that explicitly use dtype= (GH#7808) Doug Davis

• Add Coiled to list of paid support organizations (GH#7811) Kristopher Overholt

• Small tweaks to the HTML repr for Layer & HighLevelGraph (GH#7812) Genevieve Buckley

• Add dark mode support to HLG HTML repr (GH#7809) Jacob Tomlinson

• Remove compatibility entries for old distributed (GH#7801) Elliott Sales de Andrade

• Implementation of HTML repr for HighLevelGraph layers (GH#7763) Genevieve Buckley

• Update default blockwise token to avoid DataFrame column name clash (GH#6546) James Bourbeau

• Use dispatch concat for merge_asof (GH#7806) Julia Signell

• Fix upstream freq tests (GH#7795) Julia Signell

• Use more context managers from the standard library (GH#7796) James Bourbeau

• Simplify skips in parquet tests (GH#7802) Elliott Sales de Andrade

• Remove check for outdated bokeh (GH#7804) Elliott Sales de Andrade

• More test coverage uploads (GH#7799) James Bourbeau

• Remove ImportError catching from dask/__init__.py (GH#7797) James Bourbeau

• Allow DataFrame.join() to take a list of DataFrames to merge with (GH#7578) Krishan Bhasin

• Fix maximum recursion depth exception in dask.array.linspace (GH#7667) Daniel Mesejo-León

• Fix docs links (GH#7794) Julia Signell

• Initial da.select() implementation and test (GH#7760) Gabriel Miretti

• Layers must implement get_output_keys method (GH#7790) Genevieve Buckley

• Don’t include or expect freq in divisions (GH#7785) Julia Signell

• A HighLevelGraph abstract layer for map_overlap (GH#7595) Genevieve Buckley

• Always include kwarg name in drop (GH#7784) Julia Signell

• Only rechunk for median if needed (GH#7782) Julia Signell

• Add add_(prefix|suffix) to DataFrame and Series (GH#7745) tsuga

• Move read_hdf to Blockwise (GH#7625) Richard (Rick) Zamora

• Make Layer.get_output_keys officially an abstract method (GH#7775) Genevieve Buckley

• Non-dask-arrays and broadcasting in ravel_multi_index (GH#7594) Gabe Joseph

• Fix for paths ending with “/” in parquet overwrite (GH#7773) Martin Durant

• Fixing calling .visualize() with filename=None (GH#7740) Freyam Mehta

• Generate unique names for SubgraphCallable (GH#7637) Bruce Merry

• Pin fsspec to 2021.5.0 in CI (GH#7771) James Bourbeau

• Evaluate graph lazily if meta is provided in from_delayed (GH#7769) Florian Jetter

• Add meta support for DatetimeTZDtype (GH#7627) gerrymanoim

• Add dispatch label to automatic PR labeler (GH#7701) James Bourbeau

• Fix HDFS tests (GH#7752) Julia Signell

2021.06.0

Released on June 4, 2021

• Remove abstract tokens from graph keys in rewrite_blockwise (GH#7721) Richard (Rick) Zamora

• Ensure correct column order in csv project_columns (GH#7761) Richard (Rick) Zamora

• Renamed inner loop variables to avoid duplication (GH#7741) Boaz Mohar

• Do not return delayed object from to_zarr (GH#7738) Chris Roat

• Array: correct number of outputs in apply_gufunc (GH#7669) Gabe Joseph

• Rewrite da.fromfunction with da.blockwise (GH#7704) John A Kirkham

• Rename make_meta_util to make_meta (GH#7743) GALI PREM SAGAR

• Repartition before shuffle if the requested partitions are less than input partitions (GH#7715) Vibhu Jawa

• Blockwise: handle constant key inputs (GH#7734) Mads R. B. Kristensen

• Added raise to apply_gufunc (GH#7744) Boaz Mohar

• Show failing tests summary in CI (GH#7735) Genevieve Buckley

• sizeof sets in Python 3.9 (GH#7739) Mads R. B. Kristensen

• Warn if using pandas datetimelike string in dataframe.__getitem__ (GH#7749) Julia Signell

• Highlight the client.dashboard_link (GH#7747) Genevieve Buckley

• Easier link for subscribing to the Google calendar (GH#7733) Genevieve Buckley

• Automatically show graph visualization in Jupyter notebooks (GH#7716) Genevieve Buckley

• Add autofunction for unify_chunks in API docs (GH#7730) James Bourbeau

2021.05.1

Released on May 28, 2021

• Pandas compatibility (GH#7712) Julia Signell

• Fix optimize_dataframe_getitem bug (GH#7698) Richard (Rick) Zamora

• Update make_meta import in docs (GH#7713) Benjamin Zaitlen

• Implement da.searchsorted (GH#7696) Tom White

• Fix format string in error message (GH#7706) Jiaming Yuan

• Fix read_sql_table returning wrong result for single column loads (GH#7572) c-thiel

• Add slack join link in support.rst (GH#7679) Naty Clementi

• Remove unused alphabet variable (GH#7700) James Bourbeau

• Fix meta creation incase of object (GH#7586) GALI PREM SAGAR

• Add dispatch for union_categoricals (GH#7699) GALI PREM SAGAR

• Consolidate array Dispatch objects (GH#7505) James Bourbeau

• Move DataFrame dispatch.registers to their own file (GH#7503) Julia Signell

• Fix delayed with dataclasses where init=False (GH#7656) Julia Signell

• Allow a column to be named divisions (GH#7605) Julia Signell

• Stack nd array with unknown chunks (GH#7562) Chris Roat

• Promote the 2021 Dask User Survey (GH#7694) Genevieve Buckley

• Fix typo in DataFrame.set_index() (GH#7691) James Lamb

• Cleanup array API reference links (GH#7684) David Hoese

• Accept axis tuple for flip to be consistent with NumPy (GH#7675) Andrew Champion

• Bump pre-commit hook versions (GH#7676) James Bourbeau

• Cleanup to_zarr docstring (GH#7683) David Hoese

• Fix the docstring of read_orc (GH#7678) keewis

• Doc ipyparallel & mpi4py concurrent.futures (GH#7665) John A Kirkham

• Update tests to support CuPy 9 (GH#7671) Peter Andreas Entschev

• Fix some HighLevelGraph documentation inaccuracies (GH#7662) Mads R. B. Kristensen

• Fix spelling in Series getitem error message (GH#7659) Maisie Marshall

2021.05.0

Released on May 14, 2021

• Remove deprecated kind kwarg to comply with pandas 1.3.0 (GH#7653) Julia Signell

• Fix bug in DataFrame column projection (GH#7645) Richard (Rick) Zamora

• Merge global annotations when packing (GH#7565) Mads R. B. Kristensen

• Avoid inplace= in pandas set_categories (GH#7633) James Bourbeau

• Change the active-fusion default to False for Dask-Dataframe (GH#7620) Richard (Rick) Zamora

• Array: remove extraneous code from RandomState (GH#7487) Gabe Joseph

• Implement str.concat when others=None (GH#7623) Daniel Mesejo-León

• Fix dask.dataframe in sandboxed environments (GH#7601) Noah D. Brenowitz

• Support for cupyx.scipy.linalg (GH#7563) Benjamin Zaitlen

• Move timeseries and daily-stock to Blockwise (GH#7615) Richard (Rick) Zamora

• Fix bugs in broadcast join (GH#7617) Richard (Rick) Zamora

• Use Blockwise for DataFrame IO (parquet, csv, and orc) (GH#7415) Richard (Rick) Zamora

• Adding chunk & type information to Dask HighLevelGraph s (GH#7309) Genevieve Buckley

• Add pyarrow sphinx intersphinx_mapping (GH#7612) Ray Bell

• Remove skip on test freq (GH#7608) Julia Signell

• Defaults in read_parquet parameters (GH#7567) Ray Bell

• Remove ignore_abc_warning (GH#7606) Julia Signell

• Harden DataFrame merge between column-selection and index (GH#7575) Richard (Rick) Zamora

• Get rid of ignore_abc decorator (GH#7604) Julia Signell

• Remove kwarg validation for bokeh (GH#7597) Julia Signell

• Add loky example (GH#7590) Naty Clementi

• Delayed: nout when arguments become tasks (GH#7593) Gabe Joseph

• Update distributed version in mindep CI build (GH#7602) James Bourbeau

• Support all or no overlap between partition columns and real columns (GH#7541) Richard (Rick) Zamora

2021.04.1

Released on April 23, 2021

• Handle Blockwise HLG pack/unpack for concatenate=True (GH#7455) Richard (Rick) Zamora

• map_partitions: use tokenized info as name of the SubgraphCallable (GH#7524) Mads R. B. Kristensen

• Using tmp_path and tmpdir to avoid temporary files and directories hanging in the repo (GH#7592) Naty Clementi

• Contributing to docs (development guide) (GH#7591) Naty Clementi

• Add more packages to Python 3.9 CI build (GH#7588) James Bourbeau

• Array: Fix NEP-18 dispatching in finalize (GH#7508) Gabe Joseph

• Misc fixes for numpydoc (GH#7569) Matthias Bussonnier

• Avoid pandas level= keyword deprecation (GH#7577) James Bourbeau

• Map e.g. .repartition(freq="M") to .repartition(freq="MS") (GH#7504) Ruben van de Geer

• Remove hash seeding in parallel CI runs (GH#7128) Elliott Sales de Andrade

• Add defaults in parameters in to_parquet (GH#7564) Ray Bell

• Simplify transpose axes cleanup (GH#7561) Julia Signell

• Make ValueError in len(index_names) > 1 explicit it’s using fastparquet (GH#7556) Ray Bell

• Fix dict-column appending for pyarrow parquet engines (GH#7527) Richard (Rick) Zamora

• Add a documentation auto label (GH#7560) Doug Davis

• Add dask.delayed.Delayed to docs so it can be referenced by other sphinx docs (GH#7559) Doug Davis

• Fix upstream idxmaxmin for uneven split_every (GH#7538) Julia Signell

• Make normalize_token for pandas Series/DataFrame future proof (no direct block access) (GH#7318) Joris Van den Bossche

• Redesigned __setitem__ implementation (GH#7393) David Hassell

• histogram, histogramdd improvements (docs; return consistencies) (GH#7520) Doug Davis

• Force nightly pyarrow in the upstream build (GH#7530) Joris Van den Bossche

• Fix Configuration Reference (GH#7533) Benjamin Zaitlen

• Use .to_parquet on dask.dataframe in doc string (GH#7528) Ray Bell

• Avoid double msgpack serialization of HLGs (GH#7525) Mads R. B. Kristensen

• Encourage usage of yaml.safe_load() in configuration doc (GH#7529) Hristo Georgiev

• Fix reshape bug. Add relevant test. Fixes #7171. (GH#7523) JSKenyon

• Support custom_metadata= argument in to_parquet (GH#7359) Richard (Rick) Zamora

• Clean some documentation warnings (GH#7518) Daniel Mesejo-León

• Getting rid of more docs warnings (GH#7426) Julia Signell

• Added product (alias of prod) (GH#7517) Freyam Mehta

• Fix upstream __array_ufunc__ tests (GH#7494) Julia Signell

• Escape from map_overlap to map_blocks if depth is zero (GH#7481) Genevieve Buckley

• Add check_type to array assert_eq (GH#7491) Julia Signell

2021.04.0

Released on April 2, 2021

• Adding support for multidimensional histograms with dask.array.histogramdd (GH#7387) Doug Davis

• Update docs on number of threads and workers in default LocalCluster (GH#7497) cameron16

• Add labels automatically when certain files are touched in a PR (GH#7506) Julia Signell

• Extract ignore_order from kwargs (GH#7500) GALI PREM SAGAR

• Only provide installation instructions when distributed is missing (GH#7498) Matthew Rocklin

• Start adding isort (GH#7370) Julia Signell

• Add ignore_order parameter in dd.concat (GH#7473) Daniel Mesejo-León

• Use powers-of-two when displaying RAM (GH#7484) Guido Imperiale

• Added License Classifier (GH#7485) Tom Augspurger

• Replace conda with mamba (GH#7227) Guido Imperiale

• Fix typo in array docs (GH#7478) James Lamb

• Use concurrent.futures in local scheduler (GH#6322) John A Kirkham

2021.03.1

Released on March 26, 2021

• Add a dispatch for is_categorical_dtype to handle non-pandas objects (GH#7469) brandon-b-miller

• Use multiprocessing.Pool in test_read_text (GH#7472) John A Kirkham

• Add missing meta kwarg to gufunc class (GH#7423) Peter Andreas Entschev

• Example for memory-mapped Dask array (GH#7380) Dieter Weber

• Fix NumPy upstream failures xfail pandas and fastparquet failures (GH#7441) Julia Signell

• Fix bug in repartition with freq (GH#7357) Ruben van de Geer

• Fix __array_function__ dispatching for tril/triu (GH#7457) Peter Andreas Entschev

• Use concurrent.futures.Executors in a few tests (GH#7429) John A Kirkham

• Require NumPy >=1.16 (GH#7383) Guido Imperiale

• Minor sort_values housekeeping (GH#7462) Ryan Williams

• Ensure natural sort order in parquet part paths (GH#7249) Ryan Williams

• Remove global env mutation upon running test_config.py (GH#7464) Hristo Georgiev

• Update NumPy intersphinx URL (GH#7460) Gabe Joseph

• Add rot90 (GH#7440) Trevor Manz

• Update docs for required package for endpoint (GH#7454) Nick Vazquez

• Master -> main in slice_array docstring (GH#7453) Gabe Joseph

• Expand dask.utils.is_arraylike docstring (GH#7445) Doug Davis

• Simplify BlockwiseIODeps importing (GH#7420) Richard (Rick) Zamora

• Update layer annotation packing method (GH#7430) James Bourbeau

• Drop duplicate test in test_describe_empty (GH#7431) John A Kirkham

• Add Series.dot method to dataframe module (GH#7236) Madhu94

• Added df kurtosis-method and testing (GH#7273) Jan Borchmann

• Avoid quadratic-time performance for HLG culling (GH#7403) Bruce Merry

• Temporarily skip problematic sparse test (GH#7421) James Bourbeau

• Update some CI workflow names (GH#7422) James Bourbeau

• Fix HDFS test (GH#7418) Julia Signell

• Make changelog subtitles match the hierarchy (GH#7419) Julia Signell

• Add support for normalize in value_counts (GH#7342) Julia Signell

• Avoid unnecessary imports for HLG Layer unpacking and materialization (GH#7381) Richard (Rick) Zamora

• Bincount fix slicing (GH#7391) Genevieve Buckley

• Add sliding_window_view (GH#7234) Deepak Cherian

• Fix typo in docs/source/develop.rst (GH#7414) Hristo Georgiev

• Switch documentation builds for PRs to readthedocs (GH#7397) James Bourbeau

• Adds sort_values to dask.DataFrame (GH#7286) gerrymanoim

• Pin sqlalchemy<1.4.0 in CI (GH#7405) James Bourbeau

• Comment fixes (GH#7215) Ryan Williams

• Dead code removal / fixes (GH#7388) Ryan Williams

• Use single thread for pa.Table.from_pandas calls (GH#7347) Richard (Rick) Zamora

• Replace 'container' with 'image' (GH#7389) James Lamb

• DOC hyperlink repartition (GH#7394) Ray Bell

• Pass delimiter to fsspec in bag.read_text (GH#7349) Martin Durant

• Update read_hdf default mode to "r" (GH#7039) rs9w33

• Embed literals in SubgraphCallable when packing Blockwise (GH#7353) Mads R. B. Kristensen

• Update test_hdf.py to not reuse file handlers (GH#7044) rs9w33

• Require additional dependencies: cloudpickle, partd, fsspec, toolz (GH#7345) Julia Signell

• Prepare Blockwise + IO infrastructure (GH#7281) Richard (Rick) Zamora

• Remove duplicated imports from test_slicing.py (GH#7365) Hristo Georgiev

• Add test deps for pip development (GH#7360) Julia Signell

• Support int slicing for non-NumPy arrays (GH#7364) Peter Andreas Entschev

• Automatically cancel previous CI builds (GH#7348) James Bourbeau

• dask.array.asarray should handle case where xarray class is in top-level namespace (GH#7335) Tom White

• HighLevelGraph length without materializing layers (GH#7274) Gabe Joseph

• Drop support for Python 3.6 (GH#7006) James Bourbeau

• Fix fsspec usage in create_metadata_file (GH#7295) Richard (Rick) Zamora

• Change default branch from master to main (GH#7198) Julia Signell

• Add Xarray to CI software environment (GH#7338) James Bourbeau

• Update repartition argument name in error text (GH#7336) Eoin Shanaghy

• Run upstream tests based on commit message (GH#7329) James Bourbeau

• Use pytest.register_assert_rewrite on util modules (GH#7278) Bruce Merry

• Add example on using specific chunk sizes in from_array() (GH#7330) James Lamb

• Move NumPy skip into test (GH#7247) Julia Signell

2021.03.0

Released on March 5, 2021

Note

This is the first release with support for Python 3.9 and the last release with support for Python 3.6

• Bump minimum version of distributed (GH#7328) James Bourbeau

• Fix percentiles_summary with dask_cudf (GH#7325) Peter Andreas Entschev

• Temporarily revert recent Array.__setitem__ updates (GH#7326) James Bourbeau

• Blockwise.clone (GH#7312) Guido Imperiale

• NEP-35 duck array update (GH#7321) James Bourbeau

• Don’t allow setting .name for array (GH#7222) Julia Signell

• Use nearest interpolation for creating percentiles of integer input (GH#7305) Kyle Barron

• Test exp with CuPy arrays (GH#7322) John A Kirkham

• Check that computed chunks have right size and dtype (GH#7277) Bruce Merry

• pytest.mark.flaky (GH#7319) Guido Imperiale

• Contributing docs: add note to pull the latest git tags before pip installing Dask (GH#7308) Genevieve Buckley

• Support for Python 3.9 (GH#7289) Guido Imperiale

• Add broadcast-based merge implementation (GH#7143) Richard (Rick) Zamora

• Add split_every to graph_manipulation (GH#7282) Guido Imperiale

• Typo in optimize docs (GH#7306) Julius Busecke

• dask.graph_manipulation support for xarray.Dataset (GH#7276) Guido Imperiale

• Add plot width and height support for Bokeh 2.3.0 (GH#7297) James Bourbeau

• Add NumPy functions tri, triu_indices, triu_indices_from, tril_indices, tril_indices_from (GH#6997) Illviljan

• Remove “cleanup” task in DataFrame on-disk shuffle (GH#7260) Sinclair Target

• Use development version of distributed in CI (GH#7279) James Bourbeau

• Moving high level graph pack/unpack Dask (GH#7179) Mads R. B. Kristensen

• Improve performance of merge_percentiles (GH#7172) Ashwin Srinath

• DOC: add dask-sql and fugue (GH#7129) Ray Bell

• Example for working with categoricals and parquet (GH#7085) McToel

• Adds tree reduction to bincount (GH#7183) Thomas J. Fan

• Improve documentation of name in from_array (GH#7264) Bruce Merry

• Fix cumsum for empty partitions (GH#7230) Julia Signell

• Add map_blocks example to dask array creation docs (GH#7221) Julia Signell

• Fix performance issue in dask.graph_manipulation.wait_on() (GH#7258) Guido Imperiale

• Replace coveralls with codecov.io (GH#7246) Guido Imperiale

• Pin to a particular black rev in pre-commit (GH#7256) Julia Signell

• Minor typo in documentation: array-chunks.rst (GH#7254) Magnus Nord

• Fix bugs in Blockwise and ShuffleLayer (GH#7213) Richard (Rick) Zamora

• Fix parquet filtering bug for "pyarrow-dataset" with pyarrow-3.0.0 (GH#7200) Richard (Rick) Zamora

• graph_manipulation without NumPy (GH#7243) Guido Imperiale

• Support for NEP-35 (GH#6738) Peter Andreas Entschev

• Avoid running unit tests during doctest CI build (GH#7240) James Bourbeau

• Run doctests on CI (GH#7238) Julia Signell

• Cleanup code quality on set arithmetics (GH#7196) Guido Imperiale

• Add dask.array.delete (GH#7125) Julia Signell

• Unpin graphviz now that new conda-forge recipe is built (GH#7235) Julia Signell

• Don’t use NumPy 1.20 from conda-forge on Mac (GH#7211) Guido Imperiale

• map_overlap: Don’t rechunk axes without overlap (GH#7233) Deepak Cherian

• Pin graphviz to avoid issue with latest conda-forge build (GH#7232) Julia Signell

• Use html_css_files in docs for custom CSS (GH#7220) James Bourbeau

• Graph manipulation: clone, bind, checkpoint, wait_on (GH#7109) Guido Imperiale

• Fix handling of filter expressions in parquet pyarrow-dataset engine (GH#7186) Joris Van den Bossche

• Extend __setitem__ to more closely match numpy (GH#7033) David Hassell

• Clean up Python 2 syntax (GH#7195) Guido Imperiale

• Fix regression in Delayed._length (GH#7194) Guido Imperiale

• __dask_layers__() tests and tweaks (GH#7177) Guido Imperiale

• Properly convert HighLevelGraph in multiprocessing scheduler (GH#7191) Jim Crist-Harif

• Don’t fail fast in CI (GH#7188) James Bourbeau

2021.02.0

Released on February 5, 2021

• Add percentile support for NEP-35 (GH#7162) Peter Andreas Entschev

• Added support for Float64 in column assignment (GH#7173) Nils Braun

• Coarsen rechunking error (GH#7127) Davis Bennett

• Fix upstream CI tests (GH#6896) Julia Signell

• Revise HighLevelGraph Mapping API (GH#7160) Guido Imperiale

• Update low-level graph spec to use any hashable for keys (GH#7163) James Bourbeau

• Generically rebuild a collection with different keys (GH#7142) Guido Imperiale

• Make easier to link issues in PRs (GH#7130) Ray Bell

• Add dask.array.append (GH#7146) D-Stacks

• Allow dask.array.ravel to accept array_like argument (GH#7138) D-Stacks

• Fixes link in array design doc (GH#7152) Thomas J. Fan

• Fix example of using blockwise for an outer product (GH#7119) Bruce Merry

• Deprecate HighlevelGraph.dicts in favor of .layers (GH#7145) Amit Kumar

• Align FastParquetEngine with pyarrow engines (GH#7091) Richard (Rick) Zamora

• Merge annotations (GH#7102) Ian Rose

• Simplify contents of parts list in read_parquet (GH#7066) Richard (Rick) Zamora

• check_meta(): use __class__ when checking DataFrame types (GH#7099) Mads R. B. Kristensen

• Cache several properties (GH#7104) Illviljan

• Fix parquet getitem optimization (GH#7106) Richard (Rick) Zamora

• Add cytoolz back to CI environment (GH#7103) James Bourbeau

2021.01.1

Released on January 22, 2021

• Partially fix cumprod (GH#7089) Julia Signell

• Test pandas 1.1.x / 1.2.0 releases and pandas nightly (GH#6996) Joris Van den Bossche

• Use assign to avoid SettingWithCopyWarning (GH#7092) Julia Signell

• 'mode' argument passed to bokeh.output_file() (GH#7034) (GH#7075) patquem

• Skip empty partitions when doing groupby.value_counts (GH#7073) Julia Signell

• Add error messages to assert_eq() (GH#7083) James Lamb

• Make cached properties read-only (GH#7077) Illviljan

2021.01.0

Released on January 15, 2021

• map_partitions with review comments (GH#6776) Kumar Bharath Prabhu

• Make sure that population is a real list (GH#7027) Julia Signell

• Propagate storage_options in read_csv (GH#7074) Richard (Rick) Zamora

• Remove all BlockwiseIO code (GH#7067) Richard (Rick) Zamora

• Fix CI (GH#7069) James Bourbeau

• Add option to control rechunking in reshape (GH#6753) Tom Augspurger

• Fix linalg.lstsq for complex inputs (GH#7056) Johnnie Gray

• Add compression='infer' default to read_csv (GH#6960) Richard (Rick) Zamora

• Revert parameter changes in svd_compressed #7003 (GH#7004) Eric Czech

• Skip failing s3 test (GH#7064) Martin Durant

• Revert BlockwiseIO (GH#7048) Richard (Rick) Zamora

• Add some cross-references to DataFrame.to_bag() and Series.to_bag() (GH#7049) Rob Malouf

• Rewrite matmul as blockwise without contraction/concatenate (GH#7000) Rafal Wojdyla

• Use functools.cached_property in da.shape (GH#7023) Illviljan

• Use meta value in series non_empty (GH#6976) Julia Signell

• Revert “Temporarly pin sphinx version to 3.3.1 (GH#7002)” (GH#7014) Rafal Wojdyla

• Revert python-graphviz pinning (GH#7037) Julia Signell

• Accidentally committed print statement (GH#7038) Julia Signell

• Pass dropna and observed in agg (GH#6992) Julia Signell

• Add index to meta after .str.split with expand (GH#7026) Ruben van de Geer

• CI: test pyarrow 2.0 and nightly (GH#7030) Joris Van den Bossche

• Temporarily pin python-graphviz in CI (GH#7031) James Bourbeau

• Underline section in numpydoc (GH#7013) Matthias Bussonnier

• Keep normal optimizations when adding custom optimizations (GH#7016) Matthew Rocklin

• Temporarily pin sphinx version to 3.3.1 (GH#7002) Rafal Wojdyla

• DOC: Misc formatting (GH#6998) Matthias Bussonnier

• Add inline_array option to from_array (GH#6773) Tom Augspurger

• Revert “Initial pass at blockwise array creation routines (GH#6931)” (:pr:`6995) James Bourbeau

• Set npartitions in set_index (GH#6978) Julia Signell

• Upstream config serialization and inheritance (GH#6987) Jacob Tomlinson

• Bump the minimum time in test_minimum_time (GH#6988) Martin Durant

• Fix pandas dtype inference for read_parquet (GH#6985) Richard (Rick) Zamora

• Avoid data loss in set_index with sorted=True (GH#6980) Richard (Rick) Zamora

• Bugfix in read_parquet for handling un-named indices with index=False (GH#6969) Richard (Rick) Zamora

• Use __class__ when comparing meta data (GH#6981) Mads R. B. Kristensen

• Comparing string versions won’t always work (GH#6979) Rafal Wojdyla

• Fix GH#6925 (GH#6982) sdementen

• Initial pass at blockwise array creation routines (GH#6931) Ian Rose

• Simplify has_parallel_type() (GH#6927) Mads R. B. Kristensen

• Handle annotation unpacking in BlockwiseIO (GH#6934) Simon Perkins

• Avoid deprecated yield_fixture in test_sql.py (GH#6968) Richard (Rick) Zamora

• Remove bad graph logic in BlockwiseIO (GH#6933) Richard (Rick) Zamora

• Get config item if variable is None (GH#6862) Jacob Tomlinson

• Update from_pandas docstring (GH#6957) Richard (Rick) Zamora

• Prevent fuse_roots from clobbering annotations (GH#6955) Simon Perkins

2020.12.0

Released on December 10, 2020

Highlights

• Switched to CalVer for versioning scheme.

• Introduced new APIs for HighLevelGraph to enable sending high-level representations of task graphs to the distributed scheduler.

• Introduced new HighLevelGraph layer objects including BasicLayer, Blockwise, BlockwiseIO, ShuffleLayer, and more.

• Added support for applying custom Layer-level annotations like priority, retries, etc. with the dask.annotations context manager.

• Updated minimum supported version of pandas to 0.25.0 and NumPy to 1.15.1.

• Support for the pyarrow.dataset API to read_parquet.

• Several fixes to Dask Array’s SVD.

All changes

• Make observed kwarg optional (GH#6952) Julia Signell

• Min supported pandas 0.25.0 numpy 1.15.1 (GH#6895) Julia Signell

• Make order of categoricals unambiguous (GH#6949) Julia Signell

• Improve “pyarrow-dataset” statistics performance for read_parquet (GH#6918) Richard (Rick) Zamora

• Add observed keyword to groupby (GH#6854) Julia Signell

• Make sure include_path_column works when there are multiple partitions per file (GH#6911) Julia Signell

• Fix: array.overlap and array.map_overlap block sizes are incorrect when depth is an unsigned bit type (GH#6909) GFleishman

• Fix syntax error in HLG docs example (GH#6946) Mark

• Return a Bag from sample (GH#6941) Shang Wang

• Add ravel_multi_index (GH#6939) Illviljan

• Enable parquet metadata collection in parallel (GH#6921) Richard (Rick) Zamora

• Avoid using _file in progressbar if it is None (GH#6938) Mark Harfouche

• Add Zarr to upstream CI build (GH#6932) James Bourbeau

• Introduce BlockwiseIO layer (GH#6878) Richard (Rick) Zamora

• Transmit Layer Annotations to Scheduler (GH#6889) Simon Perkins

• Update opportunistic caching page to remove experimental warning (GH#6926) Timost

• Allow pyarrow >2.0.0 (GH#6772) Richard (Rick) Zamora

• Support pyarrow.dataset API for read_parquet (GH#6534) Richard (Rick) Zamora

• Add more informative error message to da.coarsen when coarsening factors do not divide shape (GH#6908) Davis Bennett

• Only run the cron CI on dask/dask not forks (GH#6905) Jacob Tomlinson

• Add annotations to ShuffleLayers (GH#6913) Matthew Rocklin

• Temporarily xfail test_from_s3 (GH#6915) James Bourbeau

• Added dataframe skew method (GH#6881) Jan Borchmann

• Fix dtype in array meta (GH#6893) Julia Signell

• Missing name arg in helm install ... (GH#6903) Ruben van de Geer

• Fix: exception when reading an item with filters (GH#6901) Martin Durant

• Add support for cupyx sparse to dask.array.dot (GH#6846) Akira Naruse

• Pin array mindeps up a bit to get the tests to pass [test-mindeps] (GH#6894) Julia Signell

• Update/remove pandas and numpy in mindeps (GH#6888) Julia Signell

• Fix ArrowEngine bug in use of clear_known_categories (GH#6887) Richard (Rick) Zamora

• Fix documentation about task scheduler (GH#6879) Zhengnan Zhao

• Add human relative time formatting utility (GH#6883) Jacob Tomlinson

• Possible fix for 6864 set_index issue (GH#6866) Richard (Rick) Zamora

• BasicLayer: remove dependency arguments (GH#6859) Mads R. B. Kristensen

• Serialization of Blockwise (GH#6848) Mads R. B. Kristensen

• Address columns=[] bug (GH#6871) Richard (Rick) Zamora

• Avoid duplicate parquet schema communication (GH#6841) Richard (Rick) Zamora

• Add create_metadata_file utility for existing parquet datasets (GH#6851) Richard (Rick) Zamora

• Improve ordering for workloads with a common terminus (GH#6779) Tom Augspurger

• Stringify utilities (GH#6852) Mads R. B. Kristensen

• Add keyword overwrite=True to to_parquet to remove dangling files when overwriting a pyarrow Dataset. (GH#6825) Greg Hayes

• Removed map_tasks() and map_basic_layers() (GH#6853) Mads R. B. Kristensen

• Introduce QR iteration to svd_compressed (GH#6813) RogerMoens

• __dask_distributed_pack__() now takes a client argument (GH#6850) Mads R. B. Kristensen

• Use map_partitions instead of delayed in set_index (GH#6837) Mads R. B. Kristensen

• Add doc hit for as_completed().update(futures) (GH#6817) manuels

• Bump GHA setup-miniconda version (GH#6847) Jacob Tomlinson

• Remove nans when setting sorted index (GH#6829) Rockwell Weiner

• Fix transpose of u in SVD (GH#6799) RogerMoens

• Migrate to GitHub Actions (GH#6794) Jacob Tomlinson

• Fix sphinx currentmodule usage (GH#6839) James Bourbeau

• Fix minimum dependencies CI builds (GH#6838) James Bourbeau

• Avoid graph materialization during Blockwise culling (GH#6815) Richard (Rick) Zamora

• Fixed typo (GH#6834) Devanshu Desai

• Use HighLevelGraph.merge in collections_to_dsk (GH#6836) Mads R. B. Kristensen

• Respect dtype in svd compression_matrix #2849 (GH#6802) RogerMoens

• Add blocksize to task name (GH#6818) Julia Signell

• Check for all-NaN partitions (GH#6821) Rockwell Weiner

• Change “institutional” SQL doc section to point to main SQL doc (GH#6823) Martin Durant

• Fix: DataFrame.join doesn’t accept Series as other (GH#6809) David Katz

• Remove to_delayed operations from to_parquet (GH#6801) Richard (Rick) Zamora

• Layer annotation docstrings improvements (GH#6806) Simon Perkins

• Avro reader (GH#6780) Martin Durant

• Rechunk array if smallest chunk size is smaller than depth (GH#6708) Julia Signell

• Add Layer Annotations (GH#6767) Simon Perkins

• Add “view code” links to documentation (GH#6793) manuels

• Add optional IO-subgraph to Blockwise Layers (GH#6715) Richard (Rick) Zamora

• Add high level graph pack/unpack for distributed (GH#6786) Mads R. B. Kristensen

• Add missing methods of the Dataframe API (GH#6789) Stephannie Jimenez Gacha

• Add doc on managing environments (GH#6778) Martin Durant

• HLG: get_all_external_keys() (GH#6774) Mads R. B. Kristensen

• Avoid rechunking in reshape with chunksize=1 (GH#6748) Tom Augspurger

• Try to make categoricals work on join (GH#6205) Julia Signell

• Fix some minor typos and trailing whitespaces in array-slice.rst (GH#6771) Magnus Nord

• Bugfix for parquet metadata writes of empty dataframe partitions (pyarrow) (GH#6741) Callum Noble

• Document meta kwarg in map_blocks and map_overlap. (GH#6763) Peter Andreas Entschev

• Begin experimenting with parallel prefix scan for cumsum and cumprod (GH#6675) Erik Welch

• Clarify differences in boolean indexing between dask and numpy arrays (GH#6764) Illviljan

• Efficient serialization of shuffle layers (GH#6760) James Bourbeau

• Config array optimize to skip fusion and return a HLG (GH#6751) Mads R. B. Kristensen

• Temporarily use pyarrow<2 in CI (GH#6759) James Bourbeau

• Fix meta for min/max reductions (GH#6736) Peter Andreas Entschev

• Add 2D possibility to da.linalg.lstsq - mirroring numpy (GH#6749) Pascal Bourgault

• CI: Fixed bug causing flaky test failure in pivot (GH#6752) Tom Augspurger

• Serialization of layers (GH#6693) Mads R. B. Kristensen

• Add attrs property to Series/Dataframe (GH#6742) Illviljan

• Removed Mutable Default Argument (GH#6747) Mads R. B. Kristensen

• Adjust parquet ArrowEngine to allow more easy subclass for writing (GH#6505) Joris Van den Bossche

• Add ShuffleStage HLG Layer (GH#6650) Richard (Rick) Zamora

• Handle literal in meta_from_array (GH#6731) Peter Andreas Entschev

• Do balanced rechunking even if chunks are the same (GH#6735) Chris Roat

• Fix docstring DataFrame.set_index (GH#6739) Gil Forsyth

• Ensure HighLevelGraph layers always contain Layer instances (GH#6716) James Bourbeau

• Map on HighLevelGraph Layers (GH#6689) Mads R. B. Kristensen

• Update overlap *_like function calls and CuPy tests (GH#6728) Peter Andreas Entschev

• Fixes for svd with __array_function__ (GH#6727) Peter Andreas Entschev

• Added doctest extension for documentation (GH#6397) Jim Circadian

• Minor fix to #5628 using @pentschev’s suggestion (GH#6724) John A Kirkham

• Change type of Dask array when meta type changes (GH#5628) Matthew Rocklin

• Add az (GH#6719) Ray Bell

• HLG: get_dependencies() of single keys (GH#6699) Mads R. B. Kristensen

• Revert “Revert “Use HighLevelGraph layers everywhere in collections (GH#6510)” (GH#6697)” (GH#6707) Tom Augspurger

• Allow *_like array creation functions to respect input array type (GH#6680) Genevieve Buckley

• Update dask-sphinx-theme version (GH#6700) Gil Forsyth

2.30.0 / 2020-10-06

Array

• Allow rechunk to evenly split into N chunks (GH#6420) Scott Sievert

2.29.0 / 2020-10-02

Array

• _repr_html_: color sides darker instead of drawing all the lines (GH#6683) Julia Signell

• Removes warning from nanstd and nanvar (GH#6667) Thomas J. Fan

• Get shape of output from original array - map_overlap (GH#6682) Julia Signell

• Replace np.searchsorted with bisect in indexing (GH#6669) Joachim B Haga

Bag

• Make sure subprocesses have a consistent hash for bag groupby (GH#6660) Itamar Turner-Trauring

Core

• Revert “Use HighLevelGraph layers everywhere in collections (GH#6510)” (GH#6697) Tom Augspurger

• Use pandas.testing (GH#6687) John A Kirkham

• Improve 128-bit floating-point skip in tests (GH#6676) Elliott Sales de Andrade

DataFrame

• Allow setting dataframe items using a bool dataframe (GH#6608) Julia Signell

Documentation

• Fix typo (GH#6692) garanews

• Fix a few typos (GH#6678) Pav A

2.28.0 / 2020-09-25

Array

• Partially reverted changes to Array indexing that produces large changes. This restores the behavior from Dask 2.25.0 and earlier, with a warning when large chunks are produced. A configuration option is provided to avoid creating the large chunks, see Efficiency. (GH#6665) Tom Augspurger

• Add meta to to_dask_array (GH#6651) Kyle Nicholson

• Fix GH#6631 and GH#6611 (GH#6632) Rafal Wojdyla

• Infer object in array reductions (GH#6629) Daniel Saxton

• Adding v_based flag for svd_flip (GH#6658) Eric Czech

• Fix flakey array mean (GH#6656) Sam Grayson

Core

• Removed dsk equality check from SubgraphCallable.__eq__ (GH#6666) Mads R. B. Kristensen

• Use HighLevelGraph layers everywhere in collections (GH#6510) Mads R. B. Kristensen

• Adds hash dunder method to SubgraphCallable for caching purposes (GH#6424) Andrew Fulton

• Stop writing commented out config files by default (GH#6647) Matthew Rocklin

DataFrame

• Add support for collect list aggregation via agg API (GH#6655) Madhur Tandon

• Slightly better error message (GH#6657) Julia Signell

2.27.0 / 2020-09-18

Array

• Preserve dtype in svd (GH#6643) Eric Czech

Core

• store(): create a single HLG layer (GH#6601) Mads R. B. Kristensen

• Add pre-commit CI build (GH#6645) James Bourbeau

• Update .pre-commit-config to latest black. (GH#6641) Julia Signell

• Update super usage to remove Python 2 compatibility (GH#6630) Poruri Sai Rahul

• Remove u string prefixes (GH#6633) Poruri Sai Rahul

DataFrame

• Improve error message for to_sql (GH#6638) Julia Signell

• Use empty list as categories (GH#6626) Julia Signell

Documentation

• Add autofunction to array api docs for more ufuncs (GH#6644) James Bourbeau

• Add a number of missing ufuncs to dask.array docs (GH#6642) Ralf Gommers

• Add HelmCluster docs (GH#6290) Jacob Tomlinson

2.26.0 / 2020-09-11

Array

• Backend-aware dtype inference for single-chunk svd (GH#6623) Eric Czech

• Make array.reduction docstring match for dtype (GH#6624) Martin Durant

• Set lower bound on compression level for svd_compressed using rows and cols (GH#6622) Eric Czech

• Improve SVD consistency and small array handling (GH#6616) Eric Czech

• Add svd_flip #6599 (GH#6613) Eric Czech

• Handle sequences containing dask Arrays (GH#6595) Gabe Joseph

• Avoid large chunks from getitem with lists (GH#6514) Tom Augspurger

• Eagerly slice numpy arrays in from_array (GH#6605) Deepak Cherian

• Restore ability to pickle dask arrays (GH#6594) Noah D. Brenowitz

• Add SVD support for short-and-fat arrays (GH#6591) Eric Czech

• Add simple chunk type registry and defer as appropriate to upcast types (GH#6393) Jon Thielen

• Align coarsen chunks by default (GH#6580) Deepak Cherian

• Fixup reshape on unknown dimensions and other testing fixes (GH#6578) Ryan Williams

Core

• Add validation and fixes for HighLevelGraph dependencies (GH#6588) Mads R. B. Kristensen

• Fix linting issue (GH#6598) Tom Augspurger

• Skip bokeh version 2.0.0 (GH#6572) John A Kirkham

DataFrame

• Added bytes/row calculation when using meta (GH#6585) McToel

• Handle min_count in Series.sum / prod (GH#6618) Daniel Saxton

• Update DataFrame.set_index docstring (GH#6549) Timost

• Always compute 0 and 1 quantiles during quantile calculations (GH#6564) Erik Welch

• Fix wrong path when reading empty csv file (GH#6573) Abdulelah Bin Mahfoodh

Documentation

• Doc: Troubleshooting dashboard 404 (GH#6215) Kilian Lieret

• Fixup extraConfig example (GH#6625) Tom Augspurger

• Update supported Python versions (GH#6609) Julia Signell

• Document dask/daskhub helm chart (GH#6560) Tom Augspurger

2.25.0 / 2020-08-28

Core

• Compare key hashes in subs() (GH#6559) Mads R. B. Kristensen

• Rerun with latest black release (GH#6568) James Bourbeau

• License update (GH#6554) Tom Augspurger

DataFrame

• Add gs read_parquet example (GH#6548) Ray Bell

Documentation

• Remove version from documentation page names (GH#6558) James Bourbeau

• Update kubernetes-helm.rst (GH#6523) David Sheldon

• Stop 2020 survey (GH#6547) Tom Augspurger

2.24.0 / 2020-08-22

Array

• Fix setting random seed in tests. (GH#6518) Elliott Sales de Andrade

• Support meta in apply gufunc (GH#6521) joshreback

• Replace cupy.sparse with cupyx.scipy.sparse (GH#6530) John A Kirkham

Dataframe

• Bump up tolerance for rolling tests (GH#6502) Julia Signell

• Implement DatFrame.__len__ (GH#6515) Tom Augspurger

• Infer arrow schema in to_parquet (for ArrowEngine`) (GH#6490) Richard (Rick) Zamora

• Fix parquet test when no pyarrow (GH#6524) Martin Durant

• Remove problematic filter arguments in ArrowEngine (GH#6527) Richard (Rick) Zamora

• Avoid schema validation by default in ArrowEngine (GH#6536) Richard (Rick) Zamora

Core

• Use unpack_collections in make_blockwise_graph (GH#6517) Thomas J. Fan

• Move key_split() from optimization.py to utils.py (GH#6529) Mads R. B. Kristensen

• Make tests run on moto server (GH#6528) Martin Durant

2.23.0 / 2020-08-14

Array

• Reduce np.zeros, ones, and full array size with broadcasting (GH#6491) Matthias Bussonnier

• Add missing meta= for trim in map_overlap (GH#6494) Peter Andreas Entschev

Bag

• Bag repartition partition size (GH#6371) joshreback

Core

• Scalar.__dask_layers__() to return self._name instead of self.key (GH#6507) Mads R. B. Kristensen

• Update dependencies correctly in fuse_root optimization (GH#6508) Mads R. B. Kristensen

DataFrame

• Adds items to dataframe (GH#6503) Thomas J. Fan

• Include compression in write_table call (GH#6499) Julia Signell

• Fixed warning in nonempty_series (GH#6485) Tom Augspurger

• Intelligently determine partitions based on type of first arg (GH#6479) Matthew Rocklin

• Fix pyarrow mkdirs (GH#6475) Julia Signell

• Fix duplicate parquet output in to_parquet (GH#6451) michaelnarodovitch

Documentation

• Fix documentation da.histogram (GH#6439) Roberto Panai

• Add agg nunique example (GH#6404) Ray Bell

• Fixed a few typos in the SQL docs (GH#6489) Mike McCarty

• Docs for SQLing (GH#6453) Martin Durant

2.22.0 / 2020-07-31

Array

• Compatibility for NumPy dtype deprecation (GH#6430) Tom Augspurger

Core

• Implement sizeof for some bytes-like objects (GH#6457) John A Kirkham

• HTTP error for new fsspec (GH#6446) Martin Durant

• When RecursionError is raised, return uuid from tokenize function (GH#6437) Julia Signell

• Install deps of upstream-dev packages (GH#6431) Tom Augspurger

• Use updated link in setup.cfg (GH#6426) Zhengnan Zhao

DataFrame

• Add single quotes around column names if strings (GH#6471) Gil Forsyth

• Refactor ArrowEngine for better read_parquet performance (GH#6346) Richard (Rick) Zamora

• Add tolist dispatch (GH#6444) GALI PREM SAGAR

• Compatibility with pandas 1.1.0rc0 (GH#6429) Tom Augspurger

• Multi value pivot table (GH#6428) joshreback

• Duplicate argument definitions in to_csv docstring (GH#6411) Jun Han (Johnson) Ooi

Documentation

• Add utility to docs to convert YAML config to env vars and back (GH#6472) Jacob Tomlinson

• Fix parameter server rendering (GH#6466) Scott Sievert

• Fixes broken links (GH#6403) Jim Circadian

• Complete parameter server implementation in docs (GH#6449) Scott Sievert

• Fix typo (GH#6436) Jack Xiaosong Xu

2.21.0 / 2020-07-17

Array

• Correct error message in array.routines.gradient() (GH#6417) johnomotani

• Fix blockwise concatenate for array with some dimension=1 (GH#6342) Matthias Bussonnier

Bag

• Fix bag.take example (GH#6418) Roberto Panai

Core

• Groups values in optimization pass should only be graph and keys – not an optimization + keys (GH#6409) Benjamin Zaitlen

• Call custom optimizations once, with kwargs provided (GH#6382) Clark Zinzow

• Include pickle5 for testing on Python 3.7 (GH#6379) John A Kirkham

DataFrame

• Correct typo in error message (GH#6422) Tom McTiernan

• Use pytest.warns to check for UserWarning (GH#6378) Richard (Rick) Zamora

• Parse bytes_per_chunk keyword from string (GH#6370) Matthew Rocklin

Documentation

• Numpydoc formatting (GH#6421) Matthias Bussonnier

• Unpin numpydoc following 1.1 release (GH#6407) Gil Forsyth

• Numpydoc formatting (GH#6402) Matthias Bussonnier

• Add instructions for using conda when installing code for development (GH#6399) Ray Bell

• Update visualize docstrings (GH#6383) Zhengnan Zhao

2.20.0 / 2020-07-02

Array

• Register sizeof for numpy zero-strided arrays (GH#6343) Matthias Bussonnier

• Use concatenate_lookup in concatenate (GH#6339) John A Kirkham

• Fix rechunking of arrays with some zero-length dimensions (GH#6335) Matthias Bussonnier

DataFrame

• Dispatch iloc` calls to getitem (GH#6355) Gil Forsyth

• Handle unnamed pandas RangeIndex in fastparquet engine (GH#6350) Richard (Rick) Zamora

• Preserve index when writing partitioned parquet datasets with pyarrow (GH#6282) Richard (Rick) Zamora

• Use ignore_index for pandas’ group_split_dispatch (GH#6251) Richard (Rick) Zamora

Documentation

• Add doc describing argument (GH#6318) asmith26

2.19.0 / 2020-06-19

Array

• Cast chunk sizes to python int dtype (GH#6326) Gil Forsyth

• Add shape=None to *_like() array creation functions (GH#6064) Anderson Banihirwe

Core

• Update expected error msg for protocol difference in fsspec (GH#6331) Gil Forsyth

• Fix for floats < 1 in parse_bytes (GH#6311) Gil Forsyth

• Fix exception causes all over the codebase (GH#6308) Ram Rachum

• Fix duplicated tests (GH#6303) James Lamb

• Remove unused testing function (GH#6304) James Lamb

DataFrame

• Add high-level CSV Subgraph (GH#6262) Gil Forsyth

• Fix ValueError when merging an index-only 1-partition dataframe (GH#6309) Krishan Bhasin

• Make index.map clear divisions. (GH#6285) Julia Signell

Documentation

• Add link to 2020 survey (GH#6328) Tom Augspurger

• Update bag.rst (GH#6317) Ben Shaver

2.18.1 / 2020-06-09

Array

• Don’t try to set name on full (GH#6299) Julia Signell

• Histogram: support lazy values for range/bins (another way) (GH#6252) Gabe Joseph

Core

• Fix exception causes in utils.py (GH#6302) Ram Rachum

• Improve performance of HighLevelGraph construction (GH#6293) Julia Signell

Documentation

• Now readthedocs builds unrelased features’ docstrings (GH#6295) Antonio Ercole De Luca

• Add asyncssh intersphinx mappings (GH#6298) Jacob Tomlinson

2.18.0 / 2020-06-05

Array

• Cast slicing index to dask array if same shape as original (GH#6273) Julia Signell

• Fix stack error message (GH#6268) Stephanie Gott

• full & full_like: error on non-scalar fill_value (GH#6129) Huite

• Support for multiple arrays in map_overlap (GH#6165) Eric Czech

• Pad resample divisions so that edges are counted (GH#6255) Julia Signell

Bag

• Random sampling of k elements from a dask bag #4799 (GH#6239) Antonio Ercole De Luca

DataFrame

• Add dropna, sort, and ascending to sort_values (GH#5880) Julia Signell

• Generalize from_dask_array (GH#6263) GALI PREM SAGAR

• Add derived docstring for SeriesGroupby.nunique (GH#6284) Julia Signell

• Remove NotImplementedError in resample with rule (GH#6274) Abdulelah Bin Mahfoodh

• Add dd.to_sql (GH#6038) Ryan Williams

Documentation

• Update remote data section (GH#6258) Ray Bell

2.17.2 / 2020-05-28

Core

• Re-add the complete extra (GH#6257) Jim Crist-Harif

DataFrame

• Raise error if resample isn’t going to give right answer (GH#6244) Julia Signell

2.17.1 / 2020-05-28

Array

• Empty array rechunk (GH#6233) Andrew Fulton

Core

• Make pyyaml required (GH#6250) Jim Crist-Harif

• Fix install commands from ImportError (GH#6238) Gaurav Sheni

• Remove issue template (GH#6249) Jacob Tomlinson

DataFrame

• Pass ignore_index to dd_shuffle from DataFrame.shuffle (GH#6247) Richard (Rick) Zamora

• Cope with missing HDF keys (GH#6204) Martin Durant

• Generalize describe & quantile apis (GH#5137) GALI PREM SAGAR

2.17.0 / 2020-05-26

Array

• Small improvements to da.pad (GH#6213) Mark Boer

• Return tuple if multiple outputs in dask.array.apply_gufunc, add test to check for tuple (GH#6207) Kai Mühlbauer

• Support stack with unknown chunksizes (GH#6195) swapna

Bag

• Random Choice on Bags (GH#6208) Antonio Ercole De Luca

Core

• Raise warning delayed.visualise() (GH#6216) Amol Umbarkar

• Ensure other pickle arguments work (GH#6229) John A Kirkham

• Overhaul fuse() config (GH#6198) Guido Imperiale

• Update dask.order.order to consider “next” nodes using both FIFO and LIFO (GH#5872) Erik Welch

DataFrame

• Use 0 as fill_value for more agg methods (GH#6245) Julia Signell

• Generalize rearrange_by_column_tasks and add DataFrame.shuffle (GH#6066) Richard (Rick) Zamora

• Xfail test_rolling_numba_engine for newer numba and older pandas (GH#6236) James Bourbeau

• Generalize fix_overlap (GH#6240) GALI PREM SAGAR

• Fix DataFrame.shape with no columns (GH#6237) noreentry

• Avoid shuffle when setting a presorted index with overlapping divisions (GH#6226) Krishan Bhasin

• Adjust the Parquet engine classes to allow more easily subclassing (GH#6211) Marius van Niekerk

• Fix dd.merge_asof with left_on='col' & right_index=True (GH#6192) noreentry

• Disable warning for concat (GH#6210) Tung Dang

• Move AUTO_BLOCKSIZE out of read_csv signature (GH#6214) Jim Crist-Harif

• .loc indexing with callable (GH#6185) Endre Mark Borza

• Avoid apply in _compute_sum_of_squares for groupby std agg (GH#6186) Richard (Rick) Zamora

• Minor correction to test_parquet (GH#6190) Brian Larsen

• Adhering to the passed pat for delimeter join and fix error message (GH#6194) GALI PREM SAGAR

• Skip test_to_parquet_with_get if no parquet libs available (GH#6188) Scott Sanderson

Documentation

• Added documentation for distributed.Event class (GH#6231) Nils Braun

• Doc write to remote (GH#6124) Ray Bell

2.16.0 / 2020-05-08

Array

• Fix array general-reduction name (GH#6176) Nick Evans

• Replace dim with shape in unravel_index (GH#6155) Julia Signell

• Moment: handle all elements being masked (GH#5339) Gabe Joseph

Core

• Remove Redundant string concatenations in dask code-base (GH#6137) GALI PREM SAGAR

• Upstream compat (GH#6159) Tom Augspurger

• Ensure sizeof of dict and sequences returns an integer (GH#6179) James Bourbeau

• Estimate python collection sizes with random sampling (GH#6154) Florian Jetter

• Update test upstream (GH#6146) Tom Augspurger

• Skip test for mindeps build (GH#6144) Tom Augspurger

• Switch default multiprocessing context to “spawn” (GH#4003) Itamar Turner-Trauring

• Update manifest to include dask-schema (GH#6140) Benjamin Zaitlen

DataFrame

• Harden inconsistent-schema handling in pyarrow-based read_parquet (GH#6160) Richard (Rick) Zamora

• Add compute kwargs to methods that write data to disk (GH#6056) Krishan Bhasin

• Fix issue where unique returns an index like result from backends (GH#6153) GALI PREM SAGAR

• Fix internal error in map_partitions with collections (GH#6103) Tom Augspurger

Documentation

• Add phase of computation to index TOC (GH#6157) Benjamin Zaitlen

• Remove unused imports in scheduling script (GH#6138) James Lamb

• Fix indent (GH#6147) Martin Durant

• Add Tom’s log config example (GH#6143) Martin Durant

2.15.0 / 2020-04-24

Array

• Update dask.array.from_array to warn when passed a Dask collection (GH#6122) James Bourbeau

• Un-numpy like behaviour in dask.array.pad (GH#6042) Mark Boer

• Add support for repeats=0 in da.repeat (GH#6080) James Bourbeau

Core

• Fix yaml layout for schema (GH#6132) Benjamin Zaitlen

• Configuration Reference (GH#6069) Benjamin Zaitlen

• Add configuration option to turn off task fusion (GH#6087) Matthew Rocklin

• Skip pyarrow on windows (GH#6094) Tom Augspurger

• Set limit to maximum length of fused key (GH#6057) Lucas Rademaker

• Add test against #6062 (GH#6072) Martin Durant

• Bump checkout action to v2 (GH#6065) James Bourbeau

DataFrame

• Generalize categorical calls to support cudf Categorical (GH#6113) GALI PREM SAGAR

• Avoid reading _metadata on every worker (GH#6017) Richard (Rick) Zamora

• Use group_split_dispatch and ignore_index in apply_concat_apply (GH#6119) Richard (Rick) Zamora

• Handle new (dtype) pandas metadata with pyarrow (GH#6090) Richard (Rick) Zamora

• Skip test_partition_on_cats_pyarrow if pyarrow is not installed (GH#6112) James Bourbeau

• Update DataFrame len to handle columns with the same name (GH#6111) James Bourbeau

• ArrowEngine bug fixes and test coverage (GH#6047) Richard (Rick) Zamora

• Added mode (GH#5958) Adam Lewis

Documentation

• Update “helm install” for helm 3 usage (GH#6130) JulianWgs

• Extend preload documentation (GH#6077) Matthew Rocklin

• Fixed small typo in DataFrame map_partitions() docstring (GH#6115) Eugene Huang

• Fix typo: “double” should be times, not plus (GH#6091) David Chudzicki

• Fix first line of array.random.* docs (GH#6063) Martin Durant

• Add section about Semaphore in distributed (GH#6053) Florian Jetter

2.14.0 / 2020-04-03

Array

• Added np.iscomplexobj implementation (GH#6045) Tom Augspurger

Core

• Update test_rearrange_disk_cleanup_with_exception to pass without cloudpickle installed (GH#6052) James Bourbeau

• Fixed flaky test-rearrange (GH#5977) Tom Augspurger

DataFrame

• Use _meta_nonempty for dtype casting in stack_partitions (GH#6061) mlondschien

• Fix bugs in _metadata creation and filtering in parquet ArrowEngine (GH#6023) Richard (Rick) Zamora

Documentation

• DOC: Add name caveats (GH#6040) Tom Augspurger

2.13.0 / 2020-03-25

Array

• Support dtype and other keyword arguments in da.random (GH#6030) Matthew Rocklin

• Register support for cupy sparse hstack/vstack (GH#5735) Corey J. Nolet

• Force self.name to str in dask.array (GH#6002) Chuanzhu Xu

Bag

• Set rename_fused_keys to None by default in bag.optimize (GH#6000) Lucas Rademaker

Core

• Copy dict in to_graphviz to prevent overwriting (GH#5996) JulianWgs

• Stricter pandas xfail (GH#6024) Tom Augspurger

• Fix CI failures (GH#6013) James Bourbeau

• Update toolz to 0.8.2 and use tlz (GH#5997) Ryan Grout

• Move Windows CI builds to GitHub Actions (GH#5862) James Bourbeau

DataFrame

• Improve path-related exceptions in read_hdf (GH#6032) psimaj

• Fix dtype handling in dd.concat (GH#6006) mlondschien

• Handle cudf’s leftsemi and leftanti joins (GH#6025) Richard J Zamora

• Remove unused npartitions variable in dd.from_pandas (GH#6019) Daniel Saxton

• Added shuffle to DataFrame.random_split (GH#5980) petiop

Documentation

• Fix indentation in scheduler-overview docs (GH#6022) Matthew Rocklin

• Update task graphs in optimize docs (GH#5928) Julia Signell

• Optionally get rid of intermediary boxes in visualize, and add more labels (GH#5976) Julia Signell

2.12.0 / 2020-03-06

Array

• Improve reuse of temporaries with numpy (GH#5933) Bruce Merry

• Make map_blocks with block_info produce a Blockwise (GH#5896) Bruce Merry

• Optimize make_blockwise_graph (GH#5940) Bruce Merry

• Fix axes ordering in da.tensordot (GH#5975) Gil Forsyth

• Adds empty mode to array.pad (GH#5931) Thomas J. Fan

Core

• Remove toolz.memoize dependency in dask.utils (GH#5978) Ryan Grout

• Close pool leaking subprocess (GH#5979) Tom Augspurger

• Pin numpydoc to 0.8.0 (fix double autoescape) (GH#5961) Gil Forsyth

• Register deterministic tokenization for range objects (GH#5947) James Bourbeau

• Unpin msgpack in CI (GH#5930) JAmes Bourbeau

• Ensure dot results are placed in unique files. (GH#5937) Elliott Sales de Andrade

• Add remaining optional dependencies to Travis 3.8 CI build environment (GH#5920) James Bourbeau

DataFrame

• Skip parquet getitem optimization for some keys (GH#5917) Tom Augspurger

• Add ignore_index argument to rearrange_by_column code path (GH#5973) Richard J Zamora

• Add DataFrame and Series memory_usage_per_partition methods (GH#5971) James Bourbeau

• xfail test_describe when using Pandas 0.24.2 (GH#5948) James Bourbeau

• Implement dask.dataframe.to_numeric (GH#5929) Julia Signell

• Add new error message content when columns are in a different order (GH#5927) Julia Signell

• Use shallow copy for assign operations when possible (GH#5740) Richard J Zamora

Documentation

• Changed above to below in dask.array.triu docs (GH#5984) Henrik Andersson

• Array slicing: fix typo in slice_with_int_dask_array error message (GH#5981) Gabe Joseph

• Grammar and formatting updates to docstrings (GH#5963) James Lamb

• Update develop doc with conda option (GH#5939) Ray Bell

• Update title of DataFrame extension docs (GH#5954) James Bourbeau

• Fixed typos in documentation (GH#5962) James Lamb

• Add original class or module as a kwarg on _bind_* methods (GH#5946) Julia Signell

• Add collect list example (GH#5938) Ray Bell

• Update optimization doc for python 3 (GH#5926) Julia Signell

2.11.0 / 2020-02-19

Array

• Cache result of Array.shape (GH#5916) Bruce Merry

• Improve accuracy of estimate_graph_size for rechunk (GH#5907) Bruce Merry

• Skip rechunk steps that do not alter chunking (GH#5909) Bruce Merry

• Support dtype and other kwargs in coarsen (GH#5903) Matthew Rocklin

• Push chunk override from map_blocks into blockwise (GH#5895) Bruce Merry

• Avoid using rewrite_blockwise for a singleton (GH#5890) Bruce Merry

• Optimize slices_from_chunks (GH#5891) Bruce Merry

• Avoid unnecessary __getitem__ in block() when chunks have correct dimensionality (GH#5884) Thomas Robitaille

Bag

• Add include_path option for dask.bag.read_text (GH#5836) Yifan Gu

• Fixes ValueError in delayed execution of bagged NumPy array (GH#5828) Surya Avala

Core

• CI: Pin msgpack (GH#5923) Tom Augspurger

• Rename test_inner to test_outer (GH#5922) Shiva Raisinghani

• quote should quote dicts too (GH#5905) Bruce Merry

• Register a normalizer for literal (GH#5898) Bruce Merry

• Improve layer name synthesis for non-HLGs (GH#5888) Bruce Merry

• Replace flake8 pre-commit-hook with upstream (GH#5892) Julia Signell

• Call pip as a module to avoid warnings (GH#5861) Cyril Shcherbin

• Close ThreadPool at exit (GH#5852) Tom Augspurger

• Remove dask.dataframe import in tokenization code (GH#5855) James Bourbeau

DataFrame

• Require pandas>=0.23 (GH#5883) Tom Augspurger

• Remove lambda from dataframe aggregation (GH#5901) Matthew Rocklin

• Fix exception chaining in dataframe/__init__.py (GH#5882) Ram Rachum

• Add support for reductions on empty dataframes (GH#5804) Shiva Raisinghani

• Expose sort= argument for groupby (GH#5801) Richard J Zamora

• Add df.empty property (GH#5711) rockwellw

• Use parquet read speed-ups from fastparquet.api.paths_to_cats. (GH#5821) Igor Gotlibovych

Documentation

• Deprecate doc_wraps (GH#5912) Tom Augspurger

• Update array internal design docs for HighLevelGraph era (GH#5889) Bruce Merry

• Move over dashboard connection docs (GH#5877) Matthew Rocklin

• Move prometheus docs from distributed.dask.org (GH#5876) Matthew Rocklin

• Removing duplicated DO block at the end (GH#5878) K.-Michael Aye

• map_blocks see also (GH#5874) Tom Augspurger

• More derived from (GH#5871) Julia Signell

• Fix typo (GH#5866) Yetunde Dada

• Fix typo in cloud.rst (GH#5860) Andrew Thomas

• Add note pointing to code of conduct and diversity statement (GH#5844) Matthew Rocklin

2.10.1 / 2020-01-30

• Fix Pandas 1.0 version comparison (GH#5851) Tom Augspurger

• Fix typo in distributed diagnostics documentation (GH#5841) Gerrit Holl

2.10.0 / 2020-01-28

• Support for pandas 1.0’s new BooleanDtype and StringDtype (GH#5815) Tom Augspurger

• Compatibility with pandas 1.0’s API breaking changes and deprecations (GH#5792) Tom Augspurger

• Fixed non-deterministic tokenization of some extension-array backed pandas objects (GH#5813) Tom Augspurger

• Fixed handling of dataclass class objects in collections (GH#5812) Matteo De Wint

• Fixed resampling with tz-aware dates when one of the endpoints fell in a non-existent time (GH#5807) dfonnegra

• Delay initial Zarr dataset creation until the computation occurs (GH#5797) Chris Roat

• Use parquet dataset statistics in more cases with the pyarrow engine (GH#5799) Richard J Zamora

• Fixed exception in groupby.std() when some of the keys were large integers (GH#5737) H. Thomson Comer

2.9.2 / 2020-01-16

Array

• Unify chunks in broadcast_arrays (GH#5765) Matthew Rocklin

Core

• xfail CSV encoding tests (GH#5791) Tom Augspurger

• Update order to handle empty dask graph (GH#5789) James Bourbeau

• Redo dask.order.order (GH#5646) Erik Welch

DataFrame

• Add transparent compression for on-disk shuffle with partd (GH#5786) Christian Wesp

• Fix repr for empty dataframes (GH#5781) Shiva Raisinghani

• Pandas 1.0.0RC0 compat (GH#5784) Tom Augspurger

• Remove buggy assertions (GH#5783) Tom Augspurger

• Pandas 1.0 compat (GH#5782) Tom Augspurger

• Fix bug in pyarrow-based read_parquet on partitioned datasets (GH#5777) Richard J Zamora

• Compat for pandas 1.0 (GH#5779) Tom Augspurger

• Fix groupby/mean error with with categorical index (GH#5776) Richard J Zamora

• Support empty partitions when performing cumulative aggregation (GH#5730) Matthew Rocklin

• set_index accepts single-item unnested list (GH#5760) Wes Roach

• Fixed partitioning in set index for ordered Categorical (GH#5715) Tom Augspurger

Documentation

• Note additional use case for normalize_token.register (GH#5766) Thomas A Caswell

• Update bag repartition docstring (GH#5772) Timost

• Small typos (GH#5771) Maarten Breddels

• Fix typo in Task Expectations docs (GH#5767) James Bourbeau

• Add docs section on task expectations to graph page (GH#5764) Devin Petersohn

2.9.1 / 2019-12-27

Array

• Support Array.view with dtype=None (GH#5736) Anderson Banihirwe

• Add dask.array.nanmedian (GH#5684) Deepak Cherian

Core

• xfail test_temporary_directory on Python 3.8 (GH#5734) James Bourbeau

• Add support for Python 3.8 (GH#5603) James Bourbeau

• Use id to dedupe constants in rewrite_blockwise (GH#5696) Jim Crist

DataFrame

• Raise error when converting a dask dataframe scalar to a boolean (GH#5743) James Bourbeau

• Ensure dataframe groupby-variance is greater than zero (GH#5728) Matthew Rocklin

• Fix DataFrame.__iter__ (GH#5719) Tom Augspurger

• Support Parquet filters in disjunctive normal form, like PyArrow (GH#5656) Matteo De Wint

• Auto-detect categorical columns in ArrowEngine-based read_parquet (GH#5690) Richard J Zamora

• Skip parquet getitem optimization tests if no engine found (GH#5697) James Bourbeau

• Fix independent optimization of parquet-getitem (GH#5613) Tom Augspurger

Documentation

• Update helm config doc (GH#5750) Ray Bell

• Link to examples.dask.org in several places (GH#5733) Tom Augspurger

• Add missing ” in performance report example (GH#5724) James Bourbeau

• Resolve several documentation build warnings (GH#5685) James Bourbeau

• add info on performance_report (GH#5713) Benjamin Zaitlen

• Add more docs disclaimers (GH#5710) Julia Signell

• Fix simple typo: wihout -> without (GH#5708) Tim Gates

• Update numpydoc dependency (GH#5694) James Bourbeau

2.9.0 / 2019-12-06

Array

• Fix da.std to work with NumPy arrays (GH#5681) James Bourbeau

Core

• Register sizeof functions for Numba and RMM (GH#5668) John A Kirkham

• Update meeting time (GH#5682) Tom Augspurger

DataFrame

• Modify dd.DataFrame.drop to use shallow copy (GH#5675) Richard J Zamora

• Fix bug in _get_md_row_groups (GH#5673) Richard J Zamora

• Close sqlalchemy engine after querying DB (GH#5629) Krishan Bhasin

• Allow dd.map_partitions to not enforce meta (GH#5660) Matthew Rocklin

• Generalize concat_unindexed_dataframes to support cudf-backend (GH#5659) Richard J Zamora

• Add dataframe resample methods (GH#5636) Benjamin Zaitlen

• Compute length of dataframe as length of first column (GH#5635) Matthew Rocklin

Documentation

• Doc fixup (GH#5665) James Bourbeau

• Update doc build instructions (GH#5640) James Bourbeau

• Fix ADL link (GH#5639) Ray Bell

• Add documentation build (GH#5617) James Bourbeau

2.8.1 / 2019-11-22

Array

• Use auto rechunking in da.rechunk if no value given (GH#5605) Matthew Rocklin

Core

• Add simple action to activate GH actions (GH#5619) James Bourbeau

DataFrame

• Fix “file_path_0” bug in aggregate_row_groups (GH#5627) Richard J Zamora

• Add chunksize argument to read_parquet (GH#5607) Richard J Zamora

• Change test_repartition_npartitions to support arch64 architecture (GH#5620) ossdev07

• Categories lost after groupby + agg (GH#5423) Oliver Hofkens

• Fixed relative path issue with parquet metadata file (GH#5608) Nuno Gomes Silva

• Enable gpu-backed covariance/correlation in dataframes (GH#5597) Richard J Zamora

Documentation

• Fix institutional faq and unknown doc warnings (GH#5616) James Bourbeau

• Add doc for some utils (GH#5609) Tom Augspurger

• Removes html_extra_path (GH#5614) James Bourbeau

• Fixed See Also referencence (GH#5612) Tom Augspurger

2.8.0 / 2019-11-14

Array

• Implement complete dask.array.tile function (GH#5574) Bouwe Andela

• Add median along an axis with automatic rechunking (GH#5575) Matthew Rocklin

• Allow da.asarray to chunk inputs (GH#5586) Matthew Rocklin

Bag

• Use key_split in Bag name (GH#5571) Matthew Rocklin

Core

• Switch Doctests to Py3.7 (GH#5573) Ryan Nazareth

• Relax get_colors test to adapt to new Bokeh release (GH#5576) Matthew Rocklin

• Add dask.blockwise.fuse_roots optimization (GH#5451) Matthew Rocklin

• Add sizeof implementation for small dicts (GH#5578) Matthew Rocklin

• Update fsspec, gcsfs, s3fs (GH#5588) Tom Augspurger

DataFrame

• Add dropna argument to groupby (GH#5579) Richard J Zamora

• Revert “Remove import of dask_cudf, which is now a part of cudf (GH#5568)” (GH#5590) Matthew Rocklin

Documentation

• Add best practice for dask.compute function (GH#5583) Matthew Rocklin

• Create FUNDING.yml (GH#5587) Gina Helfrich

• Add screencast for coordination primitives (GH#5593) Matthew Rocklin

• Move funding to .github repo (GH#5589) Tom Augspurger

• Update calendar link (GH#5569) Tom Augspurger

2.7.0 / 2019-11-08

This release drops support for Python 3.5

Array

• Reuse code for assert_eq util method (GH#5496) Vijayant

• Update da.array to always return a dask array (GH#5510) James Bourbeau

• Skip transpose on trivial inputs (GH#5523) Ryan Abernathey

• Avoid NumPy scalar string representation in tokenize (GH#5527) James Bourbeau

• Remove unnecessary tiledb shape constraint (GH#5545) Norman Barker

• Removes bytes from sparse array HTML repr (GH#5556) James Bourbeau

Core

• Drop Python 3.5 (GH#5528) James Bourbeau

• Update the use of fixtures in distributed tests (GH#5497) Matthew Rocklin

• Changed deprecated bokeh-port to dashboard-address (GH#5507) darindf

• Avoid updating with identical dicts in ensure_dict (GH#5501) James Bourbeau

• Test Upstream (GH#5516) Tom Augspurger

• Accelerate reverse_dict (GH#5479) Ryan Grout

• Update test_imports.sh (GH#5534) James Bourbeau

• Support cgroups limits on cpu count in multiprocess and threaded schedulers (GH#5499) Albert DeFusco

• Update minimum pyarrow version on CI (GH#5562) James Bourbeau

• Make cloudpickle optional (GH#5511) Guido Imperiale

DataFrame

• Add an example of index_col usage (GH#3072) Bruno Bonfils

• Explicitly use iloc for row indexing (GH#5500) Krishan Bhasin

• Accept dask arrays on columns assignemnt (GH#5224) Henrique Ribeiro-

• Implement unique and value_counts for SeriesGroupBy (GH#5358) Scott Sievert

• Add sizeof definition for pyarrow tables and columns (GH#5522) Richard J Zamora

• Enable row-group task partitioning in pyarrow-based read_parquet (GH#5508) Richard J Zamora

• Removes npartitions=’auto’ from dd.merge docstring (GH#5531) James Bourbeau

• Apply enforce error message shows non-overlapping columns. (GH#5530) Tom Augspurger

• Optimize meta_nonempty for repetitive dtypes (GH#5553) Petio Petrov

• Remove import of dask_cudf, which is now a part of cudf (GH#5568) Mads R. B. Kristensen

Documentation

• Make capitalization more consistent in FAQ docs (GH#5512) Matthew Rocklin

• Add CONTRIBUTING.md (GH#5513) Jacob Tomlinson

• Document optional dependencies (GH#5456) Prithvi MK

• Update helm chart docs to reflect new chart repo (GH#5539) Jacob Tomlinson

• Add Resampler to API docs (GH#5551) James Bourbeau

• Fix typo in read_sql_table (GH#5554) Eric Dill

• Add adaptive deployments screencast [skip ci] (GH#5566) Matthew Rocklin

2.6.0 / 2019-10-15

Core

• Call ensure_dict on graphs before entering toolz.merge (GH#5486) Matthew Rocklin

• Consolidating hash dispatch functions (GH#5476) Richard J Zamora

DataFrame

• Support Python 3.5 in Parquet code (GH#5491) Benjamin Zaitlen

• Avoid identity check in warn_dtype_mismatch (GH#5489) Tom Augspurger

• Enable unused groupby tests (GH#3480) Jörg Dietrich

• Remove old parquet and bcolz dataframe optimizations (GH#5484) Matthew Rocklin

• Add getitem optimization for read_parquet (GH#5453) Tom Augspurger

• Use _constructor_sliced method to determine Series type (GH#5480) Richard J Zamora

• Fix map(series) for unsorted base series index (GH#5459) Justin Waugh

• Fix KeyError with Groupby label (GH#5467) Ryan Nazareth

Documentation

• Use Zoom meeting instead of appear.in (GH#5494) Matthew Rocklin

• Added curated list of resources (GH#5460) Javad

• Update SSH docs to include SSHCluster (GH#5482) Matthew Rocklin

• Update “Why Dask?” page (GH#5473) Matthew Rocklin

• Fix typos in docstrings (GH#5469) garanews

2.5.2 / 2019-10-04

Array

• Correct chunk size logic for asymmetric overlaps (GH#5449) Ben Jeffery

• Make da.unify_chunks public API (GH#5443) Matthew Rocklin

DataFrame

• Fix dask.dataframe.fillna handling of Scalar object (GH#5463) Zhenqing Li

Documentation

• Remove boxes in Spark comparison page (GH#5445) Matthew Rocklin

• Add latest presentations (GH#5446) Javad

• Update cloud documentation (GH#5444) Matthew Rocklin

2.5.0 / 2019-09-27

Core

• Add sentinel no_default to get_dependencies task (GH#5420) James Bourbeau

• Update fsspec version (GH#5415) Matthew Rocklin

• Remove PY2 checks (GH#5400) Jim Crist

DataFrame

• Add option to not check meta in dd.from_delayed (GH#5436) Christopher J. Wright

• Fix test_timeseries_nulls_in_schema failures with pyarrow master (GH#5421) Richard J Zamora

• Reduce read_metadata output size in pyarrow/parquet (GH#5391) Richard J Zamora

• Test numeric edge case for repartition with npartitions. (GH#5433) amerkel2

• Unxfail pandas-datareader test (GH#5430) Tom Augspurger

• Add DataFrame.pop implementation (GH#5422) Matthew Rocklin

• Enable merge/set_index for cudf-based dataframes with cupy values (GH#5322) Richard J Zamora

• drop_duplicates support for positional subset parameter (GH#5410) Wes Roach

Documentation

• Add screencasts to array, bag, dataframe, delayed, futures and setup (GH#5429) (GH#5424) Matthew Rocklin

• Fix delimeter parsing documentation (GH#5428) Mahmut Bulut

• Update overview image (GH#5404) James Bourbeau

2.4.0 / 2019-09-13

Array

• Adds explicit h5py.File mode (GH#5390) James Bourbeau

• Provides method to compute unknown array chunks sizes (GH#5312) Scott Sievert

• Ignore runtime warning in Array compute_meta (GH#5356) estebanag

• Add _meta to Array.__dask_postpersist__ (GH#5353) Benoit Bovy

• Fixup da.asarray and da.asanyarray for datetime64 dtype and xarray objects (GH#5334) Stephan Hoyer

• Add shape implementation (GH#5293) Tom Augspurger

• Add chunktype to array text repr (GH#5289) James Bourbeau

• Array.random.choice: handle array-like non-arrays (GH#5283) Gabe Joseph

Core

• Remove deprecated code (GH#5401) Jim Crist

• Fix funcname when vectorized func has no __name__ (GH#5399) James Bourbeau

• Truncate funcname to avoid long key names (GH#5383) Matthew Rocklin

• Add support for numpy.vectorize in funcname (GH#5396) James Bourbeau

• Fixed HDFS upstream test (GH#5395) Tom Augspurger

• Support numbers and None in parse_bytes/timedelta (GH#5384) Matthew Rocklin

• Fix tokenizing of subindexes on memmapped numpy arrays (GH#5351) Henry Pinkard

• Upstream fixups (GH#5300) Tom Augspurger

DataFrame

• Allow pandas to cast type of statistics (GH#5402) Richard J Zamora

• Preserve index dtype after applying dd.pivot_table (GH#5385) therhaag

• Implement explode for Series and DataFrame (GH#5381) Arpit Solanki

• set_index on categorical fails with less categories than partitions (GH#5354) Oliver Hofkens

• Support output to a single CSV file (GH#5304) Hongjiu Zhang

• Add groupby().transform() (GH#5327) Oliver Hofkens

• Adding filter kwarg to pyarrow dataset call (GH#5348) Richard J Zamora

• Implement and check compression defaults for parquet (GH#5335) Sarah Bird

• Pass sqlalchemy params to delayed objects (GH#5332) Arpit Solanki

• Fixing schema handling in arrow-parquet (GH#5307) Richard J Zamora

• Add support for DF and Series groupby().idxmin/max() (GH#5273) Oliver Hofkens

• Add correlation calculation and add test (GH#5296) Benjamin Zaitlen

Documentation

• Numpy docstring standard has moved (GH#5405) Wes Roach

• Reference correct NumPy array name (GH#5403) Wes Roach

• Minor edits to Array chunk documentation (GH#5372) Scott Sievert

• Add methods to API docs (GH#5387) Tom Augspurger

• Add namespacing to configuration example (GH#5374) Matthew Rocklin

• Add get_task_stream and profile to the diagnostics page (GH#5375) Matthew Rocklin

• Add best practice to load data with Dask (GH#5369) Matthew Rocklin

• Update institutional-faq.rst (GH#5345) DomHudson

• Add threads and processes note to the best practices (GH#5340) Matthew Rocklin

• Update cuDF links (GH#5328) James Bourbeau

• Fixed small typo with parentheses placement (GH#5311) Eugene Huang

• Update link in reshape docstring (GH#5297) James Bourbeau

2.3.0 / 2019-08-16

Array

• Raise exception when from_array is given a dask array (GH#5280) David Hoese

• Avoid adjusting gufunc’s meta dtype twice (GH#5274) Peter Andreas Entschev

• Add meta= keyword to map_blocks and add test with sparse (GH#5269) Matthew Rocklin

• Add rollaxis and moveaxis (GH#4822) Tobias de Jong

• Always increment old chunk index (GH#5256) James Bourbeau

• Shuffle dask array (GH#3901) Tom Augspurger

• Fix ordering when indexing a dask array with a bool dask array (GH#5151) James Bourbeau

Bag

• Add workaround for memory leaks in bag generators (GH#5208) Marco Neumann

Core

• Set strict xfail option (GH#5220) James Bourbeau

• test-upstream (GH#5267) Tom Augspurger

• Fixed HDFS CI failure (GH#5234) Tom Augspurger

• Error nicely if no file size inferred (GH#5231) Jim Crist

• A few changes to config.set (GH#5226) Jim Crist

• Fixup black string normalization (GH#5227) Jim Crist

• Pin NumPy in windows tests (GH#5228) Jim Crist

• Ensure parquet tests are skipped if fastparquet and pyarrow not installed (GH#5217) James Bourbeau

• Add fsspec to readthedocs (GH#5207) Matthew Rocklin

• Bump NumPy and Pandas to 1.17 and 0.25 in CI test (GH#5179) John A Kirkham

DataFrame

• Fix DataFrame.query docstring (incorrect numexpr API) (GH#5271) Doug Davis

• Parquet metadata-handling improvements (GH#5218) Richard J Zamora

• Improve messaging around sorted parquet columns for index (GH#5265) Martin Durant

• Add rearrange_by_divisions and set_index support for cudf (GH#5205) Richard J Zamora

• Fix groupby.std() with integer colum names (GH#5096) Nicolas Hug

• Add Series.__iter__ (GH#5071) Blane

• Generalize hash_pandas_object to work for non-pandas backends (GH#5184) GALI PREM SAGAR

• Add rolling cov (GH#5154) Ivars Geidans

• Add columns argument in drop function (GH#5223) Henrique Ribeiro

Documentation

• Update institutional FAQ doc (GH#5277) Matthew Rocklin

• Add draft of institutional FAQ (GH#5214) Matthew Rocklin

• Make boxes for dask-spark page (GH#5249) Martin Durant

• Add motivation for shuffle docs (GH#5213) Matthew Rocklin

• Fix links and API entries for best-practices (GH#5246) Martin Durant

• Remove “bytes” (internal data ingestion) doc page (GH#5242) Martin Durant

• Redirect from our local distributed page to distributed.dask.org (GH#5248) Matthew Rocklin

• Cleanup API page (GH#5247) Matthew Rocklin

• Remove excess endlines from install docs (GH#5243) Matthew Rocklin

• Remove item list in phases of computation doc (GH#5245) Martin Durant

• Remove custom graphs from the TOC sidebar (GH#5241) Matthew Rocklin

• Remove experimental status of custom collections (GH#5236) James Bourbeau

• Adds table of contents to Why Dask? (GH#5244) James Bourbeau

• Moves bag overview to top-level bag page (GH#5240) James Bourbeau

• Remove use-cases in favor of stories.dask.org (GH#5238) Matthew Rocklin

• Removes redundant TOC information in index.rst (GH#5235) James Bourbeau

• Elevate dashboard in distributed diagnostics documentation (GH#5239) Martin Durant

• Updates “add” layer in HLG docs example (GH#5237) James Bourbeau

• Update GUFunc documentation (GH#5232) Matthew Rocklin

2.2.0 / 2019-08-01

Array

• Use da.from_array(…, asarray=False) if input follows NEP-18 (GH#5074) Matthew Rocklin

• Add missing attributes to from_array documentation (GH#5108) Peter Andreas Entschev

• Fix meta computation for some reduction functions (GH#5035) Peter Andreas Entschev

• Raise informative error in to_zarr if unknown chunks (GH#5148) James Bourbeau

• Remove invalid pad tests (GH#5122) Tom Augspurger

• Ignore NumPy warnings in compute_meta (GH#5103) Peter Andreas Entschev

• Fix kurtosis calc for single dimension input array (GH#5177) @andrethrill

• Support Numpy 1.17 in tests (GH#5192) Matthew Rocklin

Bag

• Supply pool to bag test to resolve intermittent failure (GH#5172) Tom Augspurger

Core

• Base dask on fsspec (GH#5064) (GH#5121) Martin Durant

• Various upstream compatibility fixes (GH#5056) Tom Augspurger

• Make distributed tests optional again. (GH#5128) Elliott Sales de Andrade

• Fix HDFS in dask (GH#5130) Martin Durant

• Ignore some more invalid value warnings. (GH#5140) Elliott Sales de Andrade

DataFrame

• Fix pd.MultiIndex size estimate (GH#5066) Brett Naul

• Generalizing has_known_categories (GH#5090) GALI PREM SAGAR

• Refactor Parquet engine (GH#4995) Richard J Zamora

• Add divide method to series and dataframe (GH#5094) msbrown47

• fix flaky partd test (GH#5111) Tom Augspurger

• Adjust is_dataframe_like to adjust for value_counts change (GH#5143) Tom Augspurger

• Generalize rolling windows to support non-Pandas dataframes (GH#5149) Nick Becker

• Avoid unnecessary aggregation in pivot_table (GH#5173) Daniel Saxton

• Add column names to apply_and_enforce error message (GH#5180) Matthew Rocklin

• Add schema keyword argument to to_parquet (GH#5150) Sarah Bird

• Remove recursion error in accessors (GH#5182) Jim Crist

• Allow fastparquet to handle gather_statistics=False for file lists (GH#5157) Richard J Zamora

Documentation

• Adds NumFOCUS badge to the README (GH#5086) James Bourbeau

• Update developer docs [ci skip] (GH#5093) Jim Crist

• Document DataFrame.set_index computataion behavior Natalya Rapstine

• Use pip install . instead of calling setup.py (GH#5139) Matthias Bussonier

• Close user survey (GH#5147) Tom Augspurger

• Fix Google Calendar meeting link (GH#5155) Loïc Estève

• Add docker image customization example (GH#5171) James Bourbeau

• Update remote-data-services after fsspec (GH#5170) Martin Durant

• Fix typo in spark.rst (GH#5164) Xavier Holt

• Update setup/python docs for async/await API (GH#5163) Matthew Rocklin

• Update Local Storage HPC documentation (GH#5165) Matthew Rocklin

2.1.0 / 2019-07-08

Array

• Add recompute= keyword to svd_compressed for lower-memory use (GH#5041) Matthew Rocklin

• Change __array_function__ implementation for backwards compatibility (GH#5043) Ralf Gommers

• Added dtype and shape kwargs to apply_along_axis (GH#3742) Davis Bennett

• Fix reduction with empty tuple axis (GH#5025) Peter Andreas Entschev

• Drop size 0 arrays in stack (GH#4978) John A Kirkham

Core

• Removes index keyword from pandas to_parquet call (GH#5075) James Bourbeau

• Fixes upstream dev CI build installation (GH#5072) James Bourbeau

• Ensure scalar arrays are not rendered to SVG (GH#5058) Willi Rath

• Environment creation overhaul (GH#5038) Tom Augspurger

• s3fs, moto compatibility (GH#5033) Tom Augspurger

• pytest 5.0 compat (GH#5027) Tom Augspurger

DataFrame

• Fix compute_meta recursion in blockwise (GH#5048) Peter Andreas Entschev

• Remove hard dependency on pandas in get_dummies (GH#5057) GALI PREM SAGAR

• Check dtypes unchanged when using DataFrame.assign (GH#5047) asmith26

• Fix cumulative functions on tables with more than 1 partition (GH#5034) tshatrov

• Handle non-divisible sizes in repartition (GH#5013) George Sakkis

• Handles timestamp and preserve_index changes in pyarrow (GH#5018) Richard J Zamora

• Fix undefined meta for str.split(expand=False) (GH#5022) Brett Naul

• Removed checks used for debugging merge_asof (GH#5011) Cody Johnson

• Don’t use type when getting accessor in dataframes (GH#4992) Matthew Rocklin

• Add melt as a method of Dask DataFrame (GH#4984) Dustin Tindall

• Adds path-like support to to_hdf (GH#5003) James Bourbeau

Documentation

• Point to latest K8s setup article in JupyterHub docs (GH#5065) Sean McKenna

• Changes vizualize to visualize (GH#5061) David Brochart

• Fix from_sequence typo in delayed best practices (GH#5045) James Bourbeau

• Add user survey link to docs (GH#5026) James Bourbeau

• Fixes typo in optimization docs (GH#5015) James Bourbeau

• Update community meeting information (GH#5006) Tom Augspurger

2.0.0 / 2019-06-25

Array

• Support automatic chunking in da.indices (GH#4981) James Bourbeau

• Err if there are no arrays to stack (GH#4975) John A Kirkham

• Asymmetrical Array Overlap (GH#4863) Michael Eaton

• Dispatch concatenate where possible within dask array (GH#4669) Hameer Abbasi

• Fix tokenization of memmapped numpy arrays on different part of same file (GH#4931) Henry Pinkard

• Preserve NumPy condition in da.asarray to preserve output shape (GH#4945) Alistair Miles

• Expand foo_like_safe usage (GH#4946) Peter Andreas Entschev

• Defer order/casting einsum parameters to NumPy implementation (GH#4914) Peter Andreas Entschev

• Remove numpy warning in moment calculation (GH#4921) Matthew Rocklin

• Fix meta_from_array to support Xarray test suite (GH#4938) Matthew Rocklin

• Cache chunk boundaries for integer slicing (GH#4923) Bruce Merry

• Drop size 0 arrays in concatenate (GH#4167) John A Kirkham

• Raise ValueError if concatenate is given no arrays (GH#4927) John A Kirkham

• Promote types in concatenate using _meta (GH#4925) John A Kirkham

• Add chunk type to html repr in Dask array (GH#4895) Matthew Rocklin

• Add Dask Array._meta attribute (GH#4543) Peter Andreas Entschev

  • Fix _meta slicing of flexible types (GH#4912) Peter Andreas Entschev

  • Minor meta construction cleanup in concatenate (GH#4937) Peter Andreas Entschev

  • Further relax Array meta checks for Xarray (GH#4944) Matthew Rocklin

  • Support meta= keyword in da.from_delayed (GH#4972) Matthew Rocklin

  • Concatenate meta along axis (GH#4977) John A Kirkham

  • Use meta in stack (GH#4976) John A Kirkham

  • Move blockwise_meta to more general compute_meta function (GH#4954) Matthew Rocklin

• Alias .partitions to .blocks attribute of dask arrays (GH#4853) Genevieve Buckley

• Drop outdated numpy_compat functions (GH#4850) John A Kirkham

• Allow da.eye to support arbitrary chunking sizes with chunks=’auto’ (GH#4834) Anderson Banihirwe

• Fix CI warnings in dask.array tests (GH#4805) Tom Augspurger

• Make map_blocks work with drop_axis + block_info (GH#4831) Bruce Merry

• Add SVG image and table in Array._repr_html_ (GH#4794) Matthew Rocklin

• ufunc: avoid __array_wrap__ in favor of __array_function__ (GH#4708) Peter Andreas Entschev

• Ensure trivial padding returns the original array (GH#4990) John A Kirkham

• Test da.block with 0-size arrays (GH#4991) John A Kirkham

Core

• Drop Python 2.7 (GH#4919) Jim Crist

• Quiet dependency installs in CI (GH#4960) Tom Augspurger

• Raise on warnings in tests (GH#4916) Tom Augspurger

• Add a diagnostics extra to setup.py (includes bokeh) (GH#4924) John A Kirkham

• Add newline delimter keyword to OpenFile (GH#4935) btw08

• Overload HighLevelGraphs values method (GH#4918) James Bourbeau

• Add __await__ method to Dask collections (GH#4901) Matthew Rocklin

• Also ignore AttributeErrors which may occur if snappy (not python-snappy) is installed (GH#4908) Mark Bell

• Canonicalize key names in config.rename (GH#4903) Ian Bolliger

• Bump minimum partd to 0.3.10 (GH#4890) Tom Augspurger

• Catch async def SyntaxError (GH#4836) James Bourbeau

• catch IOError in ensure_file (GH#4806) Justin Poehnelt

• Cleanup CI warnings (GH#4798) Tom Augspurger

• Move distributed’s parse and format functions to dask.utils (GH#4793) Matthew Rocklin

• Apply black formatting (GH#4983) James Bourbeau

• Package license file in wheels (GH#4988) John A Kirkham

DataFrame

• Add an optional partition_size parameter to repartition (GH#4416) George Sakkis

• merge_asof and prefix_reduction (GH#4877) Cody Johnson

• Allow dataframes to be indexed by dask arrays (GH#4882) Endre Mark Borza

• Avoid deprecated message parameter in pytest.raises (GH#4962) James Bourbeau

• Update test_to_records to test with lengths argument(GH#4515) asmith26

• Remove pandas pinning in Dataframe accessors (GH#4955) Matthew Rocklin

• Fix correlation of series with same names (GH#4934) Philipp S. Sommer

• Map Dask Series to Dask Series (GH#4872) Justin Waugh

• Warn in dd.merge on dtype warning (GH#4917) mcsoini

• Add groupby Covariance/Correlation (GH#4889) Benjamin Zaitlen

• keep index name with to_datetime (GH#4905) Ian Bolliger

• Add Parallel variance computation for dataframes (GH#4865) Ksenia Bobrova

• Add divmod implementation to arrays and dataframes (GH#4884) Henrique Ribeiro

• Add documentation for dataframe reshape methods (GH#4896) tpanza

• Avoid use of pandas.compat (GH#4881) Tom Augspurger

• Added accessor registration for Series, DataFrame, and Index (GH#4829) Tom Augspurger

• Add read_function keyword to read_json (GH#4810) Richard J Zamora

• Provide full type name in check_meta (GH#4819) Matthew Rocklin

• Correctly estimate bytes per row in read_sql_table (GH#4807) Lijo Jose

• Adding support of non-numeric data to describe() (GH#4791) Ksenia Bobrova

• Scalars for extension dtypes. (GH#4459) Tom Augspurger

• Call head before compute in dd.from_delayed (GH#4802) Matthew Rocklin

• Add support for rolling operations with larger window that partition size in DataFrames with Time-based index (GH#4796) Jorge Pessoa

• Update groupby-apply doc with warning (GH#4800) Tom Augspurger

• Change groupby-ness tests in _maybe_slice (GH#4786) Benjamin Zaitlen

• Add master best practices document (GH#4745) Matthew Rocklin

• Add document for how Dask works with GPUs (GH#4792) Matthew Rocklin

• Add cli API docs (GH#4788) James Bourbeau

• Ensure concat output has coherent dtypes (GH#4692) Guillaume Lemaitre

• Fixes pandas_datareader dependencies installation (GH#4989) James Bourbeau

• Accept pathlib.Path as pattern in read_hdf (GH#3335) Jörg Dietrich

Documentation

• Move CLI API docs to relavant pages (GH#4980) James Bourbeau

• Add to_datetime function to dataframe API docs Matthew Rocklin

• Add documentation entry for dask.array.ma.average (GH#4970) Bouwe Andela

• Add bag.read_avro to bag API docs (GH#4969) James Bourbeau

• Fix typo (GH#4968) mbarkhau

• Docs: Drop support for Python 2.7 (GH#4932) Hugo

• Remove requirement to modify changelog (GH#4915) Matthew Rocklin

• Add documentation about meta column order (GH#4887) Tom Augspurger

• Add documentation note in DataFrame.shift (GH#4886) Tom Augspurger

• Docs: Fix typo (GH#4868) Paweł Kordek

• Put do/don’t into boxes for delayed best practice docs (GH#3821) Martin Durant

• Doc fixups (GH#2528) Tom Augspurger

• Add quansight to paid support doc section (GH#4838) Martin Durant

• Add document for custom startup (GH#4833) Matthew Rocklin

• Allow utils.derive_from to accept functions, apply across array (GH#4804) Martin Durant

• Add “Avoid Large Partitions” section to best practices (GH#4808) Matthew Rocklin

• Update URL for joblib to new website hosting their doc (GH#4816) Christian Hudon

1.2.2 / 2019-05-08

Array

• Clarify regions kwarg to array.store (GH#4759) Martin Durant

• Add dtype= parameter to da.random.randint (GH#4753) Matthew Rocklin

• Use “row major” rather than “C order” in docstring (GH#4452) @asmith26

• Normalize Xarray datasets to Dask arrays (GH#4756) Matthew Rocklin

• Remove normed keyword in da.histogram (GH#4755) Matthew Rocklin

Bag

• Add key argument to Bag.distinct (GH#4423) Daniel Severo

Core

• Add core dask config file (GH#4774) Matthew Rocklin

• Add core dask config file to MANIFEST.in (GH#4780) James Bourbeau

• Enabling glob with HTTP file-system (GH#3926) Martin Durant

• HTTPFile.seek with whence=1 (GH#4751) Martin Durant

• Remove config key normalization (GH#4742) Jim Crist

DataFrame

• Remove explicit references to Pandas in dask.dataframe.groupby (GH#4778) Matthew Rocklin

• Add support for group_keys kwarg in DataFrame.groupby() (GH#4771) Brian Chu

• Describe doc (GH#4762) Martin Durant

• Remove explicit pandas check in cumulative aggregations (GH#4765) Nick Becker

• Added meta for read_json and test (GH#4588) Abhinav Ralhan

• Add test for dtype casting (GH#4760) Martin Durant

• Document alignment in map_partitions (GH#4757) Jim Crist

• Implement Series.str.split(expand=True) (GH#4744) Matthew Rocklin

Documentation

• Tweaks to develop.rst from trying to run tests (GH#4772) Christian Hudon

• Add document describing phases of computation (GH#4766) Matthew Rocklin

• Point users to Dask-Yarn from spark documentation (GH#4770) Matthew Rocklin

• Update images in delayed doc to remove labels (GH#4768) Martin Durant

• Explain intermediate storage for dask arrays (GH#4025) John A Kirkham

• Specify bash code-block in array best practices (GH#4764) James Bourbeau

• Add array best practices doc (GH#4705) Matthew Rocklin

• Update optimization docs now that cull is not automatic (GH#4752) Matthew Rocklin

1.2.1 / 2019-04-29

Array

• Fix map_blocks with block_info and broadcasting (GH#4737) Bruce Merry

• Make ‘minlength’ keyword argument optional in da.bincount (GH#4684) Genevieve Buckley

• Add support for map_blocks with no array arguments (GH#4713) Bruce Merry

• Add dask.array.trace (GH#4717) Danilo Horta

• Add sizeof support for cupy.ndarray (GH#4715) Peter Andreas Entschev

• Add name kwarg to from_zarr (GH#4663) Michael Eaton

• Add chunks=’auto’ to from_array (GH#4704) Matthew Rocklin

• Raise TypeError if dask array is given as shape for da.ones, zeros, empty or full (GH#4707) Genevieve Buckley

• Add TileDB backend (GH#4679) Isaiah Norton

Core

• Delay long list arguments (GH#4735) Matthew Rocklin

• Bump to numpy >= 1.13, pandas >= 0.21.0 (GH#4720) Jim Crist

• Remove file “test” (GH#4710) James Bourbeau

• Reenable development build, uses upstream libraries (GH#4696) Peter Andreas Entschev

• Remove assertion in HighLevelGraph constructor (GH#4699) Matthew Rocklin

DataFrame

• Change cum-aggregation last-nonnull-value algorithm (GH#4736) Nick Becker

• Fixup series-groupby-apply (GH#4738) Jim Crist

• Refactor array.percentile and dataframe.quantile to use t-digest (GH#4677) Janne Vuorela

• Allow naive concatenation of sorted dataframes (GH#4725) Matthew Rocklin

• Fix perf issue in dd.Series.isin (GH#4727) Jim Crist

• Remove hard pandas dependency for melt by using methodcaller (GH#4719) Nick Becker

• A few dataframe metadata fixes (GH#4695) Jim Crist

• Add Dataframe.replace (GH#4714) Matthew Rocklin

• Add ‘threshold’ parameter to pd.DataFrame.dropna (GH#4625) Nathan Matare

Documentation

• Add warning about derived docstrings early in the docstring (GH#4716) Matthew Rocklin

• Create dataframe best practices doc (GH#4703) Matthew Rocklin

• Uncomment dask_sphinx_theme (GH#4728) James Bourbeau

• Fix minor typo fix in a Queue/fire_and_forget example (GH#4709) Matthew Rocklin

• Update from_pandas docstring to match signature (GH#4698) James Bourbeau

1.2.0 / 2019-04-12

Array

• Fixed mean() and moment() on sparse arrays (GH#4525) Peter Andreas Entschev

• Add test for NEP-18. (GH#4675) Hameer Abbasi

• Allow None to say “no chunking” in normalize_chunks (GH#4656) Matthew Rocklin

• Fix limit value in auto_chunks (GH#4645) Matthew Rocklin

Core

• Updated diagnostic bokeh test for compatibility with bokeh>=1.1.0 (GH#4680) Philipp Rudiger

• Adjusts codecov’s target/threshold, disable patch (GH#4671) Peter Andreas Entschev

• Always start with empty http buffer, not None (GH#4673) Martin Durant

DataFrame

• Propagate index dtype and name when create dask dataframe from array (GH#4686) Henrique Ribeiro

• Fix ordering of quantiles in describe (GH#4647) gregrf

• Clean up and document rearrange_column_by_tasks (GH#4674) Matthew Rocklin

• Mark some parquet tests xfail (GH#4667) Peter Andreas Entschev

• Fix parquet breakages with arrow 0.13.0 (GH#4668) Martin Durant

• Allow sample to be False when reading CSV from a remote URL (GH#4634) Ian Rose

• Fix timezone metadata inference on parquet load (GH#4655) Martin Durant

• Use is_dataframe/index_like in dd.utils (GH#4657) Matthew Rocklin

• Add min_count parameter to groupby sum method (GH#4648) Henrique Ribeiro

• Correct quantile to handle unsorted quantiles (GH#4650) gregrf

Documentation

• Add delayed extra dependencies to install docs (GH#4660) James Bourbeau

1.1.5 / 2019-03-29

Array

• Ensure that we use the dtype keyword in normalize_chunks (GH#4646) Matthew Rocklin

Core

• Use recursive glob in LocalFileSystem (GH#4186) Brett Naul

• Avoid YAML deprecation (GH#4603)

• Fix CI and add set -e (GH#4605) James Bourbeau

• Support builtin sequence types in dask.visualize (GH#4602)

• unpack/repack orderedDict (GH#4623) Justin Poehnelt

• Add da.random.randint to API docs (GH#4628) James Bourbeau

• Add zarr to CI environment (GH#4604) James Bourbeau

• Enable codecov (GH#4631) Peter Andreas Entschev

DataFrame

• Support setting the index (GH#4565)

• DataFrame.itertuples accepts index, name kwargs (GH#4593) Dan O’Donovan

• Support non-Pandas series in dd.Series.unique (GH#4599) Benjamin Zaitlen

• Replace use of explicit type check with ._is_partition_type predicate (GH#4533)

• Remove additional pandas warnings in tests (GH#4576)

• Check object for name/dtype attributes rather than type (GH#4606)

• Fix comparison against pd.Series (GH#4613) amerkel2

• Fixing warning from setting categorical codes to floats (GH#4624) Julia Signell

• Fix renaming on index to_frame method (GH#4498) Henrique Ribeiro

• Fix divisions when joining two single-partition dataframes (GH#4636) Justin Waugh

• Warn if partitions overlap in compute_divisions (GH#4600) Brian Chu

• Give informative meta= warning (GH#4637) Matthew Rocklin

• Add informative error message to Series.__getitem__ (GH#4638) Matthew Rocklin

• Add clear exception message when using index or index_col in read_csv (GH#4651) Álvaro Abella Bascarán

Documentation

• Add documentation for custom groupby aggregations (GH#4571)

• Docs dataframe joins (GH#4569)

• Specify fork-based contributions (GH#4619) James Bourbeau

• correct to_parquet example in docs (GH#4641) Aaron Fowles

• Update and secure several references (GH#4649) Søren Fuglede Jørgensen

1.1.4 / 2019-03-08

Array

• Use mask selection in compress (GH#4548) John A Kirkham

• Use asarray in extract (GH#4549) John A Kirkham

• Use correct dtype when test concatenation. (GH#4539) Elliott Sales de Andrade

• Fix CuPy tests or properly marks as xfail (GH#4564) Peter Andreas Entschev

Core

• Fix local scheduler callback to deal with custom caching (GH#4542) Yu Feng

• Use parse_bytes in read_bytes(sample=…) (GH#4554) Matthew Rocklin

DataFrame

• Fix up groupby-standard deviation again on object dtype keys (GH#4541) Matthew Rocklin

• TST/CI: Updates for pandas 0.24.1 (GH#4551) Tom Augspurger

• Add ability to control number of unique elements in timeseries (GH#4557) Matthew Rocklin

• Add support in read_csv for parameter skiprows for other iterables (GH#4560) @JulianWgs

Documentation

• DataFrame to Array conversion and unknown chunks (GH#4516) Scott Sievert

• Add docs for random array creation (GH#4566) Matthew Rocklin

• Fix typo in docstring (GH#4572) Shyam Saladi

1.1.3 / 2019-03-01

Array

• Modify mean chunk functions to return dicts rather than arrays (GH#4513) Matthew Rocklin

• Change sparse installation in CI for NumPy/Python2 compatibility (GH#4537) Matthew Rocklin

DataFrame

• Make merge dispatchable on pandas/other dataframe types (GH#4522) Matthew Rocklin

• read_sql_table - datetime index fix and index type checking (GH#4474) Joe Corbett

• Use generalized form of index checking (is_index_like) (GH#4531) Benjamin Zaitlen

• Add tests for groupby reductions with object dtypes (GH#4535) Matthew Rocklin

• Fixes #4467 : Updates time_series for pandas deprecation (GH#4530) @HSR05

Documentation

• Add missing method to documentation index (GH#4528) Bart Broere

1.1.2 / 2019-02-25

Array

• Fix another unicode/mixed-type edge case in normalize_array (GH#4489) Marco Neumann

• Add dask.array.diagonal (GH#4431) Danilo Horta

• Call asanyarray in unify_chunks (GH#4506) Jim Crist

• Modify moment chunk functions to return dicts (GH#4519) Peter Andreas Entschev

Bag

• Don’t inline output keys in dask.bag (GH#4464) Jim Crist

• Ensure that bag.from_sequence always includes at least one partition (GH#4475) Anderson Banihirwe

• Implement out_type for bag.fold (GH#4502) Matthew Rocklin

• Remove map from bag keynames (GH#4500) Matthew Rocklin

• Avoid itertools.repeat in map_partitions (GH#4507) Matthew Rocklin

DataFrame

• Fix relative path parsing on windows when using fastparquet (GH#4445) Janne Vuorela

• Fix bug in pyarrow and hdfs (GH#4453) (GH#4455) Michał Jastrzębski

• df getitem with integer slices is not implemented (GH#4466) Jim Crist

• Replace cudf-specific code with dask-cudf import (GH#4470) Matthew Rocklin

• Avoid groupby.agg(callable) in groupby-var (GH#4482) Matthew Rocklin

• Consider uint types as numerical in check_meta (GH#4485) Marco Neumann

• Fix some typos in groupby comments (GH#4494) Daniel Saxton

• Add error message around set_index(inplace=True) (GH#4501) Matthew Rocklin

• meta_nonempty works with categorical index (GH#4505) Jim Crist

• Add module name to expected meta error message (GH#4499) Matthew Rocklin

• groupby-nunique works on empty chunk (GH#4504) Jim Crist

• Propagate index metadata if not specified (GH#4509) Jim Crist

Documentation

• Update docs to use from_zarr (GH#4472) John A Kirkham

• DOC: add section of Using Other S3-Compatible Services for remote-data-services (GH#4405) Aploium

• Fix header level of section in changelog (GH#4483) Bruce Merry

• Add quotes to pip install [skip-ci] (GH#4508) James Bourbeau

Core

• Extend started_cbs AFTER state is initialized (GH#4460) Marco Neumann

• Fix bug in HTTPFile._fetch_range with headers (GH#4479) (GH#4480) Ross Petchler

• Repeat optimize_blockwise for diamond fusion (GH#4492) Matthew Rocklin

1.1.1 / 2019-01-31

Array

• Add support for cupy.einsum (GH#4402) Johnnie Gray

• Provide byte size in chunks keyword (GH#4434) Adam Beberg

• Raise more informative error for histogram bins and range (GH#4430) James Bourbeau

DataFrame

• Lazily register more cudf functions and move to backends file (GH#4396) Matthew Rocklin

• Fix ORC tests for pyarrow 0.12.0 (GH#4413) Jim Crist

• rearrange_by_column: ensure that shuffle arg defaults to ‘disk’ if it’s None in dask.config (GH#4414) George Sakkis

• Implement filters for _read_pyarrow (GH#4415) George Sakkis

• Avoid checking against types in is_dataframe_like (GH#4418) Matthew Rocklin

• Pass username as ‘user’ when using pyarrow (GH#4438) Roma Sokolov

Delayed

• Fix DelayedAttr return value (GH#4440) Matthew Rocklin

Documentation

• Use SVG for pipeline graphic (GH#4406) John A Kirkham

• Add doctest-modules to py.test documentation (GH#4427) Daniel Severo

Core

• Work around psutil 5.5.0 not allowing pickling Process objects Janne Vuorela

1.1.0 / 2019-01-18

Array

• Fix the average function when there is a masked array (GH#4236) Damien Garaud

• Add allow_unknown_chunksizes to hstack and vstack (GH#4287) Paul Vecchio

• Fix tensordot for 27+ dimensions (GH#4304) Johnnie Gray

• Fixed block_info with axes. (GH#4301) Tom Augspurger

• Use safe_wraps for matmul (GH#4346) Mark Harfouche

• Use chunks=”auto” in array creation routines (GH#4354) Matthew Rocklin

• Fix np.matmul in dask.array.Array.__array_ufunc__ (GH#4363) Stephan Hoyer

• COMPAT: Re-enable multifield copy->view change (GH#4357) Diane Trout

• Calling np.dtype on a delayed object works (GH#4387) Jim Crist

• Rework normalize_array for numpy data (GH#4312) Marco Neumann

DataFrame

• Add fill_value support for series comparisons (GH#4250) James Bourbeau

• Add schema name in read_sql_table for empty tables (GH#4268) Mina Farid

• Adjust check for bad chunks in map_blocks (GH#4308) Tom Augspurger

• Add dask.dataframe.read_fwf (GH#4316) @slnguyen

• Use atop fusion in dask dataframe (GH#4229) Matthew Rocklin

• Use parallel_types() in from_pandas (GH#4331) Matthew Rocklin

• Change DataFrame._repr_data to method (GH#4330) Matthew Rocklin

• Install pyarrow fastparquet for Appveyor (GH#4338) Gábor Lipták

• Remove explicit pandas checks and provide cudf lazy registration (GH#4359) Matthew Rocklin

• Replace isinstance(…, pandas) with is_dataframe_like (GH#4375) Matthew Rocklin

• ENH: Support 3rd-party ExtensionArrays (GH#4379) Tom Augspurger

• Pandas 0.24.0 compat (GH#4374) Tom Augspurger

Documentation

• Fix link to ‘map_blocks’ function in array api docs (GH#4258) David Hoese

• Add a paragraph on Dask-Yarn in the cloud docs (GH#4260) Jim Crist

• Copy edit documentation (GH#4267), (GH#4263), (GH#4262), (GH#4277), (GH#4271), (GH#4279), (GH#4265), (GH#4295), (GH#4293), (GH#4296), (GH#4302), (GH#4306), (GH#4318), (GH#4314), (GH#4309), (GH#4317), (GH#4326), (GH#4325), (GH#4322), (GH#4332), (GH#4333), Miguel Farrajota

• Fix typo in code example (GH#4272) Daniel Li

• Doc: Update array-api.rst (GH#4259) (GH#4282) Prabakaran Kumaresshan

• Update hpc doc (GH#4266) Guillaume Eynard-Bontemps

• Doc: Replace from_avro with read_avro in documents (GH#4313) Prabakaran Kumaresshan

• Remove reference to “get” scheduler functions in docs (GH#4350) Matthew Rocklin

• Fix typo in docstring (GH#4376) Daniel Saxton

• Added documentation for dask.dataframe.merge (GH#4382) Jendrik Jördening

Core

• Avoid recursion in dask.core.get (GH#4219) Matthew Rocklin

• Remove verbose flag from pytest setup.cfg (GH#4281) Matthew Rocklin

• Support Pytest 4.0 by specifying marks explicitly (GH#4280) Takahiro Kojima

• Add High Level Graphs (GH#4092) Matthew Rocklin

• Fix SerializableLock locked and acquire methods (GH#4294) Stephan Hoyer

• Pin boto3 to earlier version in tests to avoid moto conflict (GH#4276) Martin Durant

• Treat None as missing in config when updating (GH#4324) Matthew Rocklin

• Update Appveyor to Python 3.6 (GH#4337) Gábor Lipták

• Use parse_bytes more liberally in dask.dataframe/bytes/bag (GH#4339) Matthew Rocklin

• Add a better error message when cloudpickle is missing (GH#4342) Mark Harfouche

• Support pool= keyword argument in threaded/multiprocessing get functions (GH#4351) Matthew Rocklin

• Allow updates from arbitrary Mappings in config.update, not only dicts. (GH#4356) Stuart Berg

• Move dask/array/top.py code to dask/blockwise.py (GH#4348) Matthew Rocklin

• Add has_parallel_type (GH#4395) Matthew Rocklin

• CI: Update Appveyor (GH#4381) Tom Augspurger

• Ignore non-readable config files (GH#4388) Jim Crist

1.0.0 / 2018-11-28

Array

• Add nancumsum/nancumprod unit tests (GH#4215) Guido Imperiale

DataFrame

• Add index to to_dask_dataframe docstring (GH#4232) James Bourbeau

• Text and fix when appending categoricals with fastparquet (GH#4245) Martin Durant

• Don’t reread metadata when passing ParquetFile to read_parquet (GH#4247) Martin Durant

Documentation

• Copy edit documentation (GH#4222) (GH#4224) (GH#4228) (GH#4231) (GH#4230) (GH#4234) (GH#4235) (GH#4254) Miguel Farrajota

• Updated doc for the new scheduler keyword (GH#4251) @milesial

Core

• Avoid a few warnings (GH#4223) Matthew Rocklin

• Remove dask.store module (GH#4221) Matthew Rocklin

• Remove AUTHORS.md Jim Crist

0.20.2 / 2018-11-15

Array

• Avoid fusing dependencies of atop reductions (GH#4207) Matthew Rocklin

Dataframe

• Improve memory footprint for dataframe correlation (GH#4193) Damien Garaud

• Add empty DataFrame check to boundary_slice (GH#4212) James Bourbeau

Documentation

• Copy edit documentation (GH#4197) (GH#4204) (GH#4198) (GH#4199) (GH#4200) (GH#4202) (GH#4209) Miguel Farrajota

• Add stats module namespace (GH#4206) James Bourbeau

• Fix link in dataframe documentation (GH#4208) James Bourbeau

0.20.1 / 2018-11-09

Array

• Only allocate the result space in wrapped_pad_func (GH#4153) John A Kirkham

• Generalize expand_pad_width to expand_pad_value (GH#4150) John A Kirkham

• Test da.pad with 2D linear_ramp case (GH#4162) John A Kirkham

• Fix import for broadcast_to. (GH#4168) samc0de

• Rewrite Dask Array’s pad to add only new chunks (GH#4152) John A Kirkham

• Validate index inputs to atop (GH#4182) Matthew Rocklin

Core

• Dask.config set and get normalize underscores and hyphens (GH#4143) James Bourbeau

• Only subs on core collections, not subclasses (GH#4159) Matthew Rocklin

• Add block_size=0 option to HTTPFileSystem. (GH#4171) Martin Durant

• Add traverse support for dataclasses (GH#4165) Armin Berres

• Avoid optimization on sharedicts without dependencies (GH#4181) Matthew Rocklin

• Update the pytest version for TravisCI (GH#4189) Damien Garaud

• Use key_split rather than funcname in visualize names (GH#4160) Matthew Rocklin

Dataframe

• Add fix for DataFrame.__setitem__ for index (GH#4151) Anderson Banihirwe

• Fix column choice when passing list of files to fastparquet (GH#4174) Martin Durant

• Pass engine_kwargs from read_sql_table to sqlalchemy (GH#4187) Damien Garaud

Documentation

• Fix documentation in Delayed best practices example that returned an empty list (GH#4147) Jonathan Fraine

• Copy edit documentation (GH#4164) (GH#4175) (GH#4185) (GH#4192) (GH#4191) (GH#4190) (GH#4180) Miguel Farrajota

• Fix typo in docstring (GH#4183) Carlos Valiente

0.20.0 / 2018-10-26

Array

• Fuse Atop operations (GH#3998), (GH#4081) Matthew Rocklin

• Support da.asanyarray on dask dataframes (GH#4080) Matthew Rocklin

• Remove unnecessary endianness check in datetime test (GH#4113) Elliott Sales de Andrade

• Set name=False in array foo_like functions (GH#4116) Matthew Rocklin

• Remove dask.array.ghost module (GH#4121) Matthew Rocklin

• Fix use of getargspec in dask array (GH#4125) Stephan Hoyer

• Adds dask.array.invert (GH#4127), (GH#4131) Anderson Banihirwe

• Raise informative error on arg-reduction on unknown chunksize (GH#4128), (GH#4135) Matthew Rocklin

• Normalize reversed slices in dask array (GH#4126) Matthew Rocklin

Bag

• Add bag.to_avro (GH#4076) Martin Durant

Core

• Pull num_workers from config.get (GH#4086), (GH#4093) James Bourbeau

• Fix invalid escape sequences with raw strings (GH#4112) Elliott Sales de Andrade

• Raise an error on the use of the get= keyword and set_options (GH#4077) Matthew Rocklin

• Add import for Azure DataLake storage, and add docs (GH#4132) Martin Durant

• Avoid collections.Mapping/Sequence (GH#4138) Matthew Rocklin

Dataframe

• Include index keyword in to_dask_dataframe (GH#4071) Matthew Rocklin

• add support for duplicate column names (GH#4087) Jan Koch

• Implement min_count for the DataFrame methods sum and prod (GH#4090) Bart Broere

• Remove pandas warnings in concat (GH#4095) Matthew Rocklin

• DataFrame.to_csv header option to only output headers in the first chunk (GH#3909) Rahul Vaidya

• Remove Series.to_parquet (GH#4104) Justin Dennison

• Avoid warnings and deprecated pandas methods (GH#4115) Matthew Rocklin

• Swap ‘old’ and ‘previous’ when reporting append error (GH#4130) Martin Durant

Documentation

• Copy edit documentation (GH#4073), (GH#4074), (GH#4094), (GH#4097), (GH#4107), (GH#4124), (GH#4133), (GH#4139) Miguel Farrajota

• Fix typo in code example (GH#4089) Antonino Ingargiola

• Add pycon 2018 presentation (GH#4102) Javad

• Quick description for gcsfs (GH#4109) Martin Durant

• Fixed typo in docstrings of read_sql_table method (GH#4114) TakaakiFuruse

• Make target directories in redirects if they don’t exist (GH#4136) Matthew Rocklin

0.19.4 / 2018-10-09

Array

• Implement apply_gufunc(..., axes=..., keepdims=...) (GH#3985) Markus Gonser

Bag

• Fix typo in datasets.make_people (GH#4069) Matthew Rocklin

Dataframe

• Added percentiles options for dask.dataframe.describe method (GH#4067) Zhenqing Li

• Add DataFrame.partitions accessor similar to Array.blocks (GH#4066) Matthew Rocklin

Core

• Pass get functions and Clients through scheduler keyword (GH#4062) Matthew Rocklin

Documentation

• Fix Typo on hpc example. (missing = in kwarg). (GH#4068) Matthias Bussonier

• Extensive copy-editing: (GH#4065), (GH#4064), (GH#4063) Miguel Farrajota

0.19.3 / 2018-10-05

Array

• Make da.RandomState extensible to other modules (GH#4041) Matthew Rocklin

• Support unknown dims in ravel no-op case (GH#4055) Jim Crist

• Add basic infrastructure for cupy (GH#4019) Matthew Rocklin

• Avoid asarray and lock arguments for from_array(getitem) (GH#4044) Matthew Rocklin

• Move local imports in corrcoef to global imports (GH#4030) John A Kirkham

• Move local indices import to global import (GH#4029) John A Kirkham

• Fix-up Dask Array’s fromfunction w.r.t. dtype and kwargs (GH#4028) John A Kirkham

• Don’t use dummy expansion for trim_internal in overlapped (GH#3964) Mark Harfouche

• Add unravel_index (GH#3958) John A Kirkham

Bag

• Sort result in Bag.frequencies (GH#4033) Matthew Rocklin

• Add support for npartitions=1 edge case in groupby (GH#4050) James Bourbeau

• Add new random dataset for people (GH#4018) Matthew Rocklin

• Improve performance of bag.read_text on small files (GH#4013) Eric Wolak

• Add bag.read_avro (GH#4000) (GH#4007) Martin Durant

Dataframe

• Added an index parameter to dask.dataframe.from_dask_array() for creating a dask DataFrame from a dask Array with a given index. (GH#3991) Tom Augspurger

• Improve sub-classability of dask dataframe (GH#4015) Matthew Rocklin

• Fix failing hdfs test [test-hdfs] (GH#4046) Jim Crist

• fuse_subgraphs works without normal fuse (GH#4042) Jim Crist

• Make path for reading many parquet files without prescan (GH#3978) Martin Durant

• Index in dd.from_dask_array (GH#3991) Tom Augspurger

• Making skiprows accept lists (GH#3975) Julia Signell

• Fail early in fastparquet read for nonexistent column (GH#3989) Martin Durant

Core

• Add support for npartitions=1 edge case in groupby (GH#4050) James Bourbeau

• Automatically wrap large arguments with dask.delayed in map_blocks/partitions (GH#4002) Matthew Rocklin

• Fuse linear chains of subgraphs (GH#3979) Jim Crist

• Make multiprocessing context configurable (GH#3763) Itamar Turner-Trauring

Documentation

• Extensive copy-editing (GH#4049), (GH#4034), (GH#4031), (GH#4020), (GH#4021), (GH#4022), (GH#4023), (GH#4016), (GH#4017), (GH#4010), (GH#3997), (GH#3996), Miguel Farrajota

• Update shuffle method selection docs (GH#4048) James Bourbeau

• Remove docs/source/examples, point to examples.dask.org (GH#4014) Matthew Rocklin

• Replace readthedocs links with dask.org (GH#4008) Matthew Rocklin

• Updates DataFrame.to_hdf docstring for returned values (GH#3992) James Bourbeau

0.19.2 / 2018-09-17

Array

• apply_gufunc implements automatic infer of functions output dtypes (GH#3936) Markus Gonser

• Fix array histogram range error when array has nans (GH#3980) James Bourbeau

• Issue 3937 follow up, int type checks. (GH#3956) Yu Feng

• from_array: add @martindurant’s explaining of how hashing is done for an array. (GH#3965) Mark Harfouche

• Support gradient with coordinate (GH#3949) Keisuke Fujii

Core

• Fix use of has_keyword with partial in Python 2.7 (GH#3966) Mark Harfouche

• Set pyarrow as default for HDFS (GH#3957) Matthew Rocklin

Documentation

• Use dask_sphinx_theme (GH#3963) Matthew Rocklin

• Use JupyterLab in Binder links from main page Matthew Rocklin

• DOC: fixed sphinx syntax (GH#3960) Tom Augspurger

0.19.1 / 2018-09-06

Array

• Don’t enforce dtype if result has no dtype (GH#3928) Matthew Rocklin

• Fix NumPy issubtype deprecation warning (GH#3939) Bruce Merry

• Fix arg reduction tokens to be unique with different arguments (GH#3955) Tobias de Jong

• Coerce numpy integers to ints in slicing code (GH#3944) Yu Feng

• Linalg.norm ndim along axis partial fix (GH#3933) Tobias de Jong

Dataframe

• Deterministic DataFrame.set_index (GH#3867) George Sakkis

• Fix divisions in read_parquet when dealing with filters #3831 #3930 (GH#3923) (GH#3931) @andrethrill

• Fixing returning type in categorical.as_known (GH#3888) Sriharsha Hatwar

• Fix DataFrame.assign for callables (GH#3919) Tom Augspurger

• Include partitions with no width in repartition (GH#3941) Matthew Rocklin

• Don’t constrict stage/k dtype in dataframe shuffle (GH#3942) Matthew Rocklin

Documentation

• DOC: Add hint on how to render task graphs horizontally (GH#3922) Uwe Korn

• Add try-now button to main landing page (GH#3924) Matthew Rocklin

0.19.0 / 2018-08-29

Array

• Support coordinate in gradient (GH#3949) Keisuke Fujii

• Fix argtopk split_every bug (GH#3810) Guido Imperiale

• Ensure result computing dask.array.isnull() always gives a numpy array (GH#3825) Stephan Hoyer

• Support concatenate for scipy.sparse in dask array (GH#3836) Matthew Rocklin

• Fix argtopk on 32-bit systems. (GH#3823) Elliott Sales de Andrade

• Normalize keys in rechunk (GH#3820) Matthew Rocklin

• Allow shape of dask.array to be a numpy array (GH#3844) Mark Harfouche

• Fix numpy deprecation warning on tuple indexing (GH#3851) Tobias de Jong

• Rename ghost module to overlap (GH#3830) Robert Sare

• Re-add the ghost import to da __init__ (GH#3861) Jim Crist

• Ensure copy preserves masked arrays (GH#3852) Tobias de Jong

DataFrame

• Added dtype and sparse keywords to dask.dataframe.get_dummies() (GH#3792) Tom Augspurger

• Added dask.dataframe.to_dask_array() for converting a Dask Series or DataFrame to a Dask Array, possibly with known chunk sizes (GH#3884) Tom Augspurger

• Changed the behavior for dask.array.asarray() for dask dataframe and series inputs. Previously, the series was eagerly converted to an in-memory NumPy array before creating a dask array with known chunks sizes. This caused unexpectedly high memory usage. Now, no intermediate NumPy array is created, and a Dask array with unknown chunk sizes is returned (GH#3884) Tom Augspurger

• DataFrame.iloc (GH#3805) Tom Augspurger

• When reading multiple paths, expand globs. (GH#3828) Irina Truong

• Added index column name after resample (GH#3833) Eric Bonfadini

• Add (lazy) shape property to dataframe and series (GH#3212) Henrique Ribeiro

• Fix failing hdfs test [test-hdfs] (GH#3858) Jim Crist

• Fixes for pyarrow 0.10.0 release (GH#3860) Jim Crist

• Rename to_csv keys for diagnostics (GH#3890) Matthew Rocklin

• Match pandas warnings for concat sort (GH#3897) Tom Augspurger

• Include filename in read_csv (GH#3908) Julia Signell

Core

• Better error message on import when missing common dependencies (GH#3771) Danilo Horta

• Drop Python 3.4 support (GH#3840) Jim Crist

• Remove expired deprecation warnings (GH#3841) Jim Crist

• Add DASK_ROOT_CONFIG environment variable (GH#3849) Joe Hamman

• Don’t cull in local scheduler, do cull in delayed (GH#3856) Jim Crist

• Increase conda download retries (GH#3857) Jim Crist

• Add python_requires and Trove classifiers (GH#3855) @hugovk

• Fix collections.abc deprecation warnings in Python 3.7.0 (GH#3876) Jan Margeta

• Allow dot jpeg to xfail in visualize tests (GH#3896) Matthew Rocklin

• Add Python 3.7 to travis.yml (GH#3894) Matthew Rocklin

• Add expand_environment_variables to dask.config (GH#3893) Joe Hamman

Docs

• Fix typo in import statement of diagnostics (GH#3826) John Mrziglod

• Add link to YARN docs (GH#3838) Jim Crist

• fix of minor typos in landing page index.html (GH#3746) Christoph Moehl

• Update delayed-custom.rst (GH#3850) Anderson Banihirwe

• DOC: clarify delayed docstring (GH#3709) Scott Sievert

• Add new presentations (GH#3880) Javad

• Add dask array normalize_chunks to documentation (GH#3878) Daniel Rothenberg

• Docs: Fix link to snakeviz (GH#3900) Hans Moritz Günther

• Add missing ` to docstring (GH#3915) @rtobar

0.18.2 / 2018-07-23

Array

• Reimplemented argtopk to make it release the GIL (GH#3610) Guido Imperiale

• Don’t overlap on non-overlapped dimensions in map_overlap (GH#3653) Matthew Rocklin

• Fix linalg.tsqr for dimensions of uncertain length (GH#3662) Jeremy Chen

• Break apart uneven array-of-int slicing to separate chunks (GH#3648) Matthew Rocklin

• Align auto chunks to provided chunks, rather than shape (GH#3679) Matthew Rocklin

• Adds endpoint and retstep support for linspace (GH#3675) James Bourbeau

• Implement .blocks accessor (GH#3689) Matthew Rocklin

• Add block_info keyword to map_blocks functions (GH#3686) Matthew Rocklin

• Slice by dask array of ints (GH#3407) Guido Imperiale

• Support dtype in arange (GH#3722) Guido Imperiale

• Fix argtopk with uneven chunks (GH#3720) Guido Imperiale

• Raise error when replace=False in da.choice (GH#3765) James Bourbeau

• Update chunks in Array.__setitem__ (GH#3767) Itamar Turner-Trauring

• Add a chunksize convenience property (GH#3777) Jacob Tomlinson

• Fix and simplify array slicing behavior when step < 0 (GH#3702) Ziyao Wei

• Ensure to_zarr with return_stored True returns a Dask Array (GH#3786) John A Kirkham

Bag

• Add last_endline optional parameter in to_textfiles (GH#3745) George Sakkis

Dataframe

• Add aggregate function for rolling objects (GH#3772) Gerome Pistre

• Properly tokenize cumulative groupby aggregations (GH#3799) Cloves Almeida

Delayed

• Add the @ operator to the delayed objects (GH#3691) Mark Harfouche

• Add delayed best practices to documentation (GH#3737) Matthew Rocklin

• Fix @delayed decorator for methods and add tests (GH#3757) Ziyao Wei

Core

• Fix extra progressbar (GH#3669) Mike Neish

• Allow tasks back onto ordering stack if they have one dependency (GH#3652) Matthew Rocklin

• Prefer end-tasks with low numbers of dependencies when ordering (GH#3588) Tom Augspurger

• Add assert_eq to top-level modules (GH#3726) Matthew Rocklin

• Test that dask collections can hold scipy.sparse arrays (GH#3738) Matthew Rocklin

• Fix setup of lz4 decompression functions (GH#3782) Elliott Sales de Andrade

• Add datasets module (GH#3780) Matthew Rocklin

0.18.1 / 2018-06-22

Array

• from_array now supports scalar types and nested lists/tuples in input, just like all numpy functions do; it also produces a simpler graph when the input is a plain ndarray (GH#3568) Guido Imperiale

• Fix slicing of big arrays due to cumsum dtype bug (GH#3620) Marco Rossi

• Add Dask Array implementation of pad (GH#3578) John A Kirkham

• Fix array random API examples (GH#3625) James Bourbeau

• Add average function to dask array (GH#3640) James Bourbeau

• Tokenize ghost_internal with axes (GH#3643) Matthew Rocklin

• Add outer for Dask Arrays (GH#3658) John A Kirkham

DataFrame

• Add Index.to_series method (GH#3613) Henrique Ribeiro

• Fix missing partition columns in pyarrow-parquet (GH#3636) Martin Durant

Core

• Minor tweaks to CI (GH#3629) Guido Imperiale

• Add back dask.utils.effective_get (GH#3642) Matthew Rocklin

• DASK_CONFIG dictates config write location (GH#3621) Jim Crist

• Replace ‘collections’ key in unpack_collections with unique key (GH#3632) Yu Feng

• Avoid deepcopy in dask.config.set (GH#3649) Matthew Rocklin

0.18.0 / 2018-06-14

Array

• Add to/from_zarr for Zarr-format datasets and arrays (GH#3460) Martin Durant

• Experimental addition of generalized ufunc support, apply_gufunc, gufunc, and as_gufunc (GH#3109) (GH#3526) (GH#3539) Markus Gonser

• Avoid unnecessary rechunking tasks (GH#3529) Matthew Rocklin

• Compute dtypes at runtime for fft (GH#3511) Matthew Rocklin

• Generate UUIDs for all da.store operations (GH#3540) Martin Durant

• Correct internal dimension of Dask’s SVD (GH#3517) John A Kirkham

• BUG: do not raise IndexError for identity slice in array.vindex (GH#3559) Scott Sievert

• Adds isneginf and isposinf (GH#3581) John A Kirkham

• Drop Dask Array’s learn module (GH#3580) John A Kirkham

• added sfqr (short-and-fat) as a counterpart to tsqr… (GH#3575) Jeremy Chen

• Allow 0-width chunks in dask.array.rechunk (GH#3591) Marc Pfister

• Document Dask Array’s nan_to_num in public API (GH#3599) John A Kirkham

• Show block example (GH#3601) John A Kirkham

• Replace token= keyword with name= in map_blocks (GH#3597) Matthew Rocklin

• Disable locking in to_zarr (needed for using to_zarr in a distributed context) (GH#3607) John A Kirkham

• Support Zarr Arrays in to_zarr/from_zarr (GH#3561) John A Kirkham

• Added recursion to array/linalg/tsqr to better manage the single core bottleneck (GH#3586) Jeremy Chan (GH#3396) Guido Imperiale

Dataframe

• Add to/read_json (GH#3494) Martin Durant

• Adds index to unsupported arguments for DataFrame.rename method (GH#3522) James Bourbeau

• Adds support to subset Dask DataFrame columns using numpy.ndarray, pandas.Series, and pandas.Index objects (GH#3536) James Bourbeau

• Raise error if meta columns do not match dataframe (GH#3485) Christopher Ren

• Add index to unsupprted argument for DataFrame.rename (GH#3522) James Bourbeau

• Adds support for subsetting DataFrames with pandas Index/Series and numpy ndarrays (GH#3536) James Bourbeau

• Dataframe sample method docstring fix (GH#3566) James Bourbeau

• fixes dd.read_json to infer file compression (GH#3594) Matt Lee

• Adds n to sample method (GH#3606) James Bourbeau

• Add fastparquet ParquetFile object support (GH#3573) @andrethrill

Bag

• Rename method= keyword to shuffle= in bag.groupby (GH#3470) Matthew Rocklin

Core

• Replace get= keyword with scheduler= keyword (GH#3448) Matthew Rocklin

• Add centralized dask.config module to handle configuration for all Dask subprojects (GH#3432) (GH#3513) (GH#3520) Matthew Rocklin

• Add dask-ssh CLI Options and Description. (GH#3476) @beomi

• Read whole files fix regardless of header for HTTP (GH#3496) Martin Durant

• Adds synchronous scheduler syntax to debugging docs (GH#3509) James Bourbeau

• Replace dask.set_options with dask.config.set (GH#3502) Matthew Rocklin

• Update sphinx readthedocs-theme (GH#3516) Matthew Rocklin

• Introduce “auto” value for normalize_chunks (GH#3507) Matthew Rocklin

• Fix check in configuration with env=None (GH#3562) Simon Perkins

• Update sizeof definitions (GH#3582) Matthew Rocklin

• Remove –verbose flag from travis-ci (GH#3477) Matthew Rocklin

• Remove “da.random” from random array keys (GH#3604) Matthew Rocklin

0.17.5 / 2018-05-16

Array

• Fix rechunk with chunksize of -1 in a dict (GH#3469) Stephan Hoyer

• einsum now accepts the split_every parameter (GH#3471) Guido Imperiale

• Improved slicing performance (GH#3479) Yu Feng

DataFrame

• Compatibility with pandas 0.23.0 (GH#3499) Tom Augspurger

0.17.4 / 2018-05-03

Dataframe

• Add support for indexing Dask DataFrames with string subclasses (GH#3461) James Bourbeau

• Allow using both sorted_index and chunksize in read_hdf (GH#3463) Pierre Bartet

• Pass filesystem to arrow piece reader (GH#3466) Martin Durant

• Switches to using dask.compat string_types (GH#3462) James Bourbeau

0.17.3 / 2018-05-02

Array

• Add einsum for Dask Arrays (GH#3412) Simon Perkins

• Add piecewise for Dask Arrays (GH#3350) John A Kirkham

• Fix handling of nan in broadcast_shapes (GH#3356) John A Kirkham

• Add isin for dask arrays (GH#3363). Stephan Hoyer

• Overhauled topk for Dask Arrays: faster algorithm, particularly for large k’s; added support for multiple axes, recursive aggregation, and an option to pick the bottom k elements instead. (GH#3395) Guido Imperiale

• The topk API has changed from topk(k, array) to the more conventional topk(array, k). The legacy API still works but is now deprecated. (GH#2965) Guido Imperiale

• New function argtopk for Dask Arrays (GH#3396) Guido Imperiale

• Fix handling partial depth and boundary in map_overlap (GH#3445) John A Kirkham

• Add gradient for Dask Arrays (GH#3434) John A Kirkham

DataFrame

• Allow t as shorthand for table in to_hdf for pandas compatibility (GH#3330) Jörg Dietrich

• Added top level isna method for Dask DataFrames (GH#3294) Christopher Ren

• Fix selection on partition column on read_parquet for engine="pyarrow" (GH#3207) Uwe Korn

• Added DataFrame.squeeze method (GH#3366) Christopher Ren

• Added infer_divisions option to read_parquet to specify whether read engines should compute divisions (GH#3387) Jon Mease

• Added support for inferring division for engine="pyarrow" (GH#3387) Jon Mease

• Provide more informative error message for meta= errors (GH#3343) Matthew Rocklin

• add orc reader (GH#3284) Martin Durant

• Default compression for parquet now always Snappy, in line with pandas (GH#3373) Martin Durant

• Fixed bug in Dask DataFrame and Series comparisons with NumPy scalars (GH#3436) James Bourbeau

• Remove outdated requirement from repartition docstring (GH#3440) Jörg Dietrich

• Fixed bug in aggregation when only a Series is selected (GH#3446) Jörg Dietrich

• Add default values to make_timeseries (GH#3421) Matthew Rocklin

Core

• Support traversing collections in persist, visualize, and optimize (GH#3410) Jim Crist

• Add schedule= keyword to compute and persist. This replaces common use of the get= keyword (GH#3448) Matthew Rocklin

0.17.2 / 2018-03-21

Array

• Add broadcast_arrays for Dask Arrays (GH#3217) John A Kirkham

• Add bitwise_* ufuncs (GH#3219) John A Kirkham

• Add optional axis argument to squeeze (GH#3261) John A Kirkham

• Validate inputs to atop (GH#3307) Matthew Rocklin

• Avoid calls to astype in concatenate if all parts have the same dtype (GH#3301) Martin Durant

DataFrame

• Fixed bug in shuffle due to aggressive truncation (GH#3201) Matthew Rocklin

• Support specifying categorical columns on read_parquet with categories=[…] for engine="pyarrow" (GH#3177) Uwe Korn

• Add dd.tseries.Resampler.agg (GH#3202) Richard Postelnik

• Support operations that mix dataframes and arrays (GH#3230) Matthew Rocklin

• Support extra Scalar and Delayed args in dd.groupby._Groupby.apply (GH#3256) Gabriele Lanaro

Bag

• Support joining against single-partitioned bags and delayed objects (GH#3254) Matthew Rocklin

Core

• Fixed bug when using unexpected but hashable types for keys (GH#3238) Daniel Collins

• Fix bug in task ordering so that we break ties consistently with the key name (GH#3271) Matthew Rocklin

• Avoid sorting tasks in order when the number of tasks is very large (GH#3298) Matthew Rocklin

0.17.1 / 2018-02-22

Array

• Corrected dimension chunking in indices (GH#3166, GH#3167) Simon Perkins

• Inline store_chunk calls for store’s return_stored option (GH#3153) John A Kirkham

• Compatibility with struct dtypes for NumPy 1.14.1 release (GH#3187) Matthew Rocklin

DataFrame

• Bugfix to allow column assignment of pandas datetimes(GH#3164) Max Epstein

Core

• New file-system for HTTP(S), allowing direct loading from specific URLs (GH#3160) Martin Durant

• Fix bug when tokenizing partials with no keywords (GH#3191) Matthew Rocklin

• Use more recent LZ4 API (GH#3157) Thrasibule

• Introduce output stream parameter for progress bar (GH#3185) Dieter Weber

0.17.0 / 2018-02-09

Array

• Added a support object-type arrays for nansum, nanmin, and nanmax (GH#3133) Keisuke Fujii

• Update error handling when len is called with empty chunks (GH#3058) Xander Johnson

• Fixes a metadata bug with store’s return_stored option (GH#3064) John A Kirkham

• Fix a bug in optimization.fuse_slice to properly handle when first input is None (GH#3076) James Bourbeau

• Support arrays with unknown chunk sizes in percentile (GH#3107) Matthew Rocklin

• Tokenize scipy.sparse arrays and np.matrix (GH#3060) Roman Yurchak

DataFrame

• Support month timedeltas in repartition(freq=…) (GH#3110) Matthew Rocklin

• Avoid mutation in dataframe groupby tests (GH#3118) Matthew Rocklin

• read_csv, read_table, and read_parquet accept iterables of paths (GH#3124) Jim Crist

• Deprecates the dd.to_delayed function in favor of the existing method (GH#3126) Jim Crist

• Return dask.arrays from df.map_partitions calls when the UDF returns a numpy array (GH#3147) Matthew Rocklin

• Change handling of columns and index in dd.read_parquet to be more consistent, especially in handling of multi-indices (GH#3149) Jim Crist

• fastparquet append=True allowed to create new dataset (GH#3097) Martin Durant

• dtype rationalization for sql queries (GH#3100) Martin Durant

Bag

• Document bag.map_paritions function may receive either a list or generator. (GH#3150) Nir

Core

• Change default task ordering to prefer nodes with few dependents and then many downstream dependencies (GH#3056) Matthew Rocklin

• Add color= option to visualize to color by task order (GH#3057) (GH#3122) Matthew Rocklin

• Deprecate dask.bytes.open_text_files (GH#3077) Jim Crist

• Remove short-circuit hdfs reads handling due to maintenance costs. May be re-added in a more robust manner later (GH#3079) Jim Crist

• Add dask.base.optimize for optimizing multiple collections without computing. (GH#3071) Jim Crist

• Rename dask.optimize module to dask.optimization (GH#3071) Jim Crist

• Change task ordering to do a full traversal (GH#3066) Matthew Rocklin

• Adds an optimize_graph keyword to all to_delayed methods to allow controlling whether optimizations occur on conversion. (GH#3126) Jim Crist

• Support using pyarrow for hdfs integration (GH#3123) Jim Crist

• Move HDFS integration and tests into dask repo (GH#3083) Jim Crist

• Remove write_bytes (GH#3116) Jim Crist

0.16.1 / 2018-01-09

Array

• Fix handling of scalar percentile values in percentile (GH#3021) James Bourbeau

• Prevent bool() coercion from calling compute (GH#2958) Albert DeFusco

• Add matmul (GH#2904) John A Kirkham

• Support N-D arrays with matmul (GH#2909) John A Kirkham

• Add vdot (GH#2910) John A Kirkham

• Explicit chunks argument for broadcast_to (GH#2943) Stephan Hoyer

• Add meshgrid (GH#2938) John A Kirkham and (GH#3001) Markus Gonser

• Preserve singleton chunks in fftshift/ifftshift (GH#2733) John A Kirkham

• Fix handling of negative indexes in vindex and raise errors for out of bounds indexes (GH#2967) Stephan Hoyer

• Add flip, flipud, fliplr (GH#2954) John A Kirkham

• Add float_power ufunc (GH#2962) (GH#2969) John A Kirkham

• Compatibility for changes to structured arrays in the upcoming NumPy 1.14 release (GH#2964) Tom Augspurger

• Add block (GH#2650) John A Kirkham

• Add frompyfunc (GH#3030) Jim Crist

• Add the return_stored option to store for chaining stored results (GH#2980) John A Kirkham

DataFrame

• Fixed naming bug in cumulative aggregations (GH#3037) Martijn Arts

• Fixed dd.read_csv when names is given but header is not set to None (GH#2976) Martijn Arts

• Fixed dd.read_csv so that passing instances of CategoricalDtype in dtype will result in known categoricals (GH#2997) Tom Augspurger

• Prevent bool() coercion from calling compute (GH#2958) Albert DeFusco

• DataFrame.read_sql() (GH#2928) to an empty database tables returns an empty dask dataframe Apostolos Vlachopoulos

• Compatibility for reading Parquet files written by PyArrow 0.8.0 (GH#2973) Tom Augspurger

• Correctly handle the column name (df.columns.name) when reading in dd.read_parquet (GH#2973) Tom Augspurger

• Fixed dd.concat losing the index dtype when the data contained a categorical (GH#2932) Tom Augspurger

• Add dd.Series.rename (GH#3027) Jim Crist

• DataFrame.merge() now supports merging on a combination of columns and the index (GH#2960) Jon Mease

• Removed the deprecated dd.rolling* methods, in preparation for their removal in the next pandas release (GH#2995) Tom Augspurger

• Fix metadata inference bug in which single-partition series were mistakenly special cased (GH#3035) Jim Crist

• Add support for Series.str.cat (GH#3028) Jim Crist

Core

• Improve 32-bit compatibility (GH#2937) Matthew Rocklin

• Change task prioritization to avoid upwards branching (GH#3017) Matthew Rocklin

0.16.0 / 2017-11-17

This is a major release. It includes breaking changes, new protocols, and a large number of bug fixes.

Array

• Add atleast_1d, atleast_2d, and atleast_3d (GH#2760) (GH#2765) John A Kirkham

• Add allclose (GH#2771) by John A Kirkham

• Remove random.different_seeds from Dask Array API docs (GH#2772) John A Kirkham

• Deprecate vnorm in favor of dask.array.linalg.norm (GH#2773) John A Kirkham

• Reimplement unique to be lazy (GH#2775) John A Kirkham

• Support broadcasting of Dask Arrays with 0-length dimensions (GH#2784) John A Kirkham

• Add asarray and asanyarray to Dask Array API docs (GH#2787) James Bourbeau

• Support unique’s return_* arguments (GH#2779) John A Kirkham

• Simplify _unique_internal (GH#2850) (GH#2855) John A Kirkham

• Avoid removing some getter calls in array optimizations (GH#2826) Jim Crist

DataFrame

• Support pyarrow in dd.to_parquet (GH#2868) Jim Crist

• Fixed DataFrame.quantile and Series.quantile returning nan when missing values are present (GH#2791) Tom Augspurger

• Fixed DataFrame.quantile losing the result .name when q is a scalar (GH#2791) Tom Augspurger

• Fixed dd.concat return a dask.Dataframe when concatenating a single series along the columns, matching pandas’ behavior (GH#2800) James Munroe

• Fixed default inplace parameter for DataFrame.eval to match the pandas defualt for pandas >= 0.21.0 (GH#2838) Tom Augspurger

• Fix exception when calling DataFrame.set_index on text column where one of the partitions was empty (GH#2831) Jesse Vogt

• Do not raise exception when calling DataFrame.set_index on empty dataframe (GH#2827) Jesse Vogt

• Fixed bug in Dataframe.fillna when filling with a Series value (GH#2810) Tom Augspurger

• Deprecate old argument ordering in dd.to_parquet to better match convention of putting the dataframe first (GH#2867) Jim Crist

• df.astype(categorical_dtype -> known categoricals (GH#2835) Jim Crist

• Test against Pandas release candidate (GH#2814) Tom Augspurger

• Add more tests for read_parquet(engine=’pyarrow’) (GH#2822) Uwe Korn

• Remove unnecessary map_partitions in aggregate (GH#2712) Christopher Prohm

• Fix bug calling sample on empty partitions (GH#2818) @xwang777

• Error nicely when parsing dates in read_csv (GH#2863) Jim Crist

• Cleanup handling of passing filesystem objects to PyArrow readers (GH#2527) @fjetter

• Support repartitioning even if there are no divisions (GH#2873) @Ced4

• Support reading/writing to hdfs using pyarrow in dd.to_parquet (GH#2894, GH#2881) Jim Crist

Core

• Allow tuples as sharedict keys (GH#2763) Matthew Rocklin

• Calling compute within a dask.distributed task defaults to distributed scheduler (GH#2762) Matthew Rocklin

• Auto-import gcsfs when gcs:// protocol is used (GH#2776) Matthew Rocklin

• Fully remove dask.async module, use dask.local instead (GH#2828) Thomas Caswell

• Compatibility with bokeh 0.12.10 (GH#2844) Tom Augspurger

• Reduce test memory usage (GH#2782) Jim Crist

• Add Dask collection interface (GH#2748) Jim Crist

• Update Dask collection interface during XArray integration (GH#2847) Matthew Rocklin

• Close resource profiler process on __exit__ (GH#2871) Jim Crist

• Fix S3 tests (GH#2875) Jim Crist

• Fix port for bokeh dashboard in docs (GH#2889) Ian Hopkinson

• Wrap Dask filesystems for PyArrow compatibility (GH#2881) Jim Crist

0.15.4 / 2017-10-06

Array

• da.random.choice now works with array arguments (GH#2781)

• Support indexing in arrays with np.int (fixes regression) (GH#2719)

• Handle zero dimension with rechunking (GH#2747)

• Support -1 as an alias for “size of the dimension” in chunks (GH#2749)

• Call mkdir in array.to_npy_stack (GH#2709)

DataFrame

• Added the .str accessor to Categoricals with string categories (GH#2743)

• Support int96 (spark) datetimes in parquet writer (GH#2711)

• Pass on file scheme to fastparquet (GH#2714)

• Support Pandas 0.21 (GH#2737)

Bag

• Add tree reduction support for foldby (GH#2710)

Core

• Drop s3fs from pip install dask[complete] (GH#2750)

0.15.3 / 2017-09-24

Array

• Add masked arrays (GH#2301)

• Add *_like array creation functions (GH#2640)

• Indexing with unsigned integer array (GH#2647)

• Improved slicing with boolean arrays of different dimensions (GH#2658)

• Support literals in top and atop (GH#2661)

• Optional axis argument in cumulative functions (GH#2664)

• Improve tests on scalars with assert_eq (GH#2681)

• Fix norm keepdims (GH#2683)

• Add ptp (GH#2691)

• Add apply_along_axis (GH#2690) and apply_over_axes (GH#2702)

DataFrame

• Added Series.str[index] (GH#2634)

• Allow the groupby by param to handle columns and index levels (GH#2636)

• DataFrame.to_csv and Bag.to_textfiles now return the filenames to

  which they have written (GH#2655)

• Fix combination of partition_on and append in to_parquet (GH#2645)

• Fix for parquet file schemes (GH#2667)

• Repartition works with mixed categoricals (GH#2676)

Core

• python setup.py test now runs tests (GH#2641)

• Added new cheatsheet (GH#2649)

• Remove resize tool in Bokeh plots (GH#2688)

0.15.2 / 2017-08-25

Array

• Remove spurious keys from map_overlap graph (GH#2520)

• where works with non-bool condition and scalar values (GH#2543) (GH#2549)

• Improve compress (GH#2541) (GH#2545) (GH#2555)

• Add argwhere, _nonzero, and where(cond) (GH#2539)

• Generalize vindex in dask.array to handle multi-dimensional indices (GH#2573)

• Add choose method (GH#2584)

• Split code into reorganized files (GH#2595)

• Add linalg.norm (GH#2597)

• Add diff, ediff1d (GH#2607), (GH#2609)

• Improve dtype inference and reflection (GH#2571)

Bag

• Remove deprecated Bag behaviors (GH#2525)

DataFrame

• Support callables in assign (GH#2513)

• better error messages for read_csv (GH#2522)

• Add dd.to_timedelta (GH#2523)

• Verify metadata in from_delayed (GH#2534) (GH#2591)

• Add DataFrame.isin (GH#2558)

• Read_hdf supports iterables of files (GH#2547)

Core

• Remove bare except: blocks everywhere (GH#2590)

0.15.1 / 2017-07-08

• Add storage_options to to_textfiles and to_csv (GH#2466)

• Rechunk and simplify rfftfreq (GH#2473), (GH#2475)

• Better support ndarray subclasses (GH#2486)

• Import star in dask.distributed (GH#2503)

• Threadsafe cache handling with tokenization (GH#2511)

0.15.0 / 2017-06-09

Array

• Add dask.array.stats submodule (GH#2269)

• Support ufunc.outer (GH#2345)

• Optimize fancy indexing by reducing graph overhead (GH#2333) (GH#2394)

• Faster array tokenization using alternative hashes (GH#2377)

• Added the matmul @ operator (GH#2349)

• Improved coverage of the numpy.fft module (GH#2320) (GH#2322) (GH#2327) (GH#2323)

• Support NumPy’s __array_ufunc__ protocol (GH#2438)

Bag

• Fix bug where reductions on bags with no partitions would fail (GH#2324)

• Add broadcasting and variadic db.map top-level function. Also remove auto-expansion of tuples as map arguments (GH#2339)

• Rename Bag.concat to Bag.flatten (GH#2402)

DataFrame

• Parquet improvements (GH#2277) (GH#2422)

Core

• Move dask.async module to dask.local (GH#2318)

• Support callbacks with nested scheduler calls (GH#2397)

• Support pathlib.Path objects as uris (GH#2310)

0.14.3 / 2017-05-05

DataFrame

• Pandas 0.20.0 support

0.14.2 / 2017-05-03

Array

• Add da.indices (GH#2268), da.tile (GH#2153), da.roll (GH#2135)

• Simultaneously support drop_axis and new_axis in da.map_blocks (GH#2264)

• Rechunk and concatenate work with unknown chunksizes (GH#2235) and (GH#2251)

• Support non-numpy container arrays, notably sparse arrays (GH#2234)

• Tensordot contracts over multiple axes (GH#2186)

• Allow delayed targets in da.store (GH#2181)

• Support interactions against lists and tuples (GH#2148)

• Constructor plugins for debugging (GH#2142)

• Multi-dimensional FFTs (single chunk) (GH#2116)

Bag

• to_dataframe enforces consistent types (GH#2199)

DataFrame

• Set_index always fully sorts the index (GH#2290)

• Support compatibility with pandas 0.20.0 (GH#2249), (GH#2248), and (GH#2246)

• Support Arrow Parquet reader (GH#2223)

• Time-based rolling windows (GH#2198)

• Repartition can now create more partitions, not just less (GH#2168)

Core

• Always use absolute paths when on POSIX file system (GH#2263)

• Support user provided graph optimizations (GH#2219)

• Refactor path handling (GH#2207)

• Improve fusion performance (GH#2129), (GH#2131), and (GH#2112)

0.14.1 / 2017-03-22

Array

• Micro-optimize optimizations (GH#2058)

• Change slicing optimizations to avoid fusing raw numpy arrays (GH#2075) (GH#2080)

• Dask.array operations now work on numpy arrays (GH#2079)

• Reshape now works in a much broader set of cases (GH#2089)

• Support deepcopy python protocol (GH#2090)

• Allow user-provided FFT implementations in da.fft (GH#2093)

DataFrame

• Fix to_parquet with empty partitions (GH#2020)

• Optional npartitions='auto' mode in set_index (GH#2025)

• Optimize shuffle performance (GH#2032)

• Support efficient repartitioning along time windows like repartition(freq='12h') (GH#2059)

• Improve speed of categorize (GH#2010)

• Support single-row dataframe arithmetic (GH#2085)

• Automatically avoid shuffle when setting index with a sorted column (GH#2091)

• Improve handling of integer-na handling in read_csv (GH#2098)

Delayed

• Repeated attribute access on delayed objects uses the same key (GH#2084)

Core

• Improve naming of nodes in dot visuals to avoid generic apply (GH#2070)

• Ensure that worker processes have different random seeds (GH#2094)

0.14.0 / 2017-02-24

Array

• Fix corner cases with zero shape and misaligned values in arange (GH#1902), (GH#1904), (GH#1935), (GH#1955), (GH#1956)

• Improve concatenation efficiency (GH#1923)

• Avoid hashing in from_array if name is provided (GH#1972)

Bag

• Repartition can now increase number of partitions (GH#1934)

• Fix bugs in some reductions with empty partitions (GH#1939), (GH#1950), (GH#1953)

DataFrame

• Support non-uniform categoricals (GH#1877), (GH#1930)

• Groupby cumulative reductions (GH#1909)

• DataFrame.loc indexing now supports lists (GH#1913)

• Improve multi-level groupbys (GH#1914)

• Improved HTML and string repr for DataFrames (GH#1637)

• Parquet append (GH#1940)

• Add dd.demo.daily_stock function for teaching (GH#1992)

Delayed

• Add traverse= keyword to delayed to optionally avoid traversing nested data structures (GH#1899)

• Support Futures in from_delayed functions (GH#1961)

• Improve serialization of decorated delayed functions (GH#1969)

Core

• Improve windows path parsing in corner cases (GH#1910)

• Rename tasks when fusing (GH#1919)

• Add top level persist function (GH#1927)

• Propagate errors= keyword in byte handling (GH#1954)

• Dask.compute traverses Python collections (GH#1975)

• Structural sharing between graphs in dask.array and dask.delayed (GH#1985)

0.13.0 / 2017-01-02

Array

• Mandatory dtypes on dask.array. All operations maintain dtype information and UDF functions like map_blocks now require a dtype= keyword if it can not be inferred. (GH#1755)

• Support arrays without known shapes, such as arises when slicing arrays with arrays or converting dataframes to arrays (GH#1838)

• Support mutation by setting one array with another (GH#1840)

• Tree reductions for covariance and correlations. (GH#1758)

• Add SerializableLock for better use with distributed scheduling (GH#1766)

• Improved atop support (GH#1800)

• Rechunk optimization (GH#1737), (GH#1827)

Bag

• Avoid wrong results when recomputing the same groupby twice (GH#1867)

DataFrame

• Add map_overlap for custom rolling operations (GH#1769)

• Add shift (GH#1773)

• Add Parquet support (GH#1782) (GH#1792) (GH#1810), (GH#1843), (GH#1859), (GH#1863)

• Add missing methods combine, abs, autocorr, sem, nsmallest, first, last, prod, (GH#1787)

• Approximate nunique (GH#1807), (GH#1824)

• Reductions with multiple output partitions (for operations like drop_duplicates) (GH#1808), (GH#1823) (GH#1828)

• Add delitem and copy to DataFrames, increasing mutation support (GH#1858)

Delayed

• Changed behaviour for delayed(nout=0) and delayed(nout=1): delayed(nout=1) does not default to out=None anymore, and delayed(nout=0) is also enabled. I.e. functions with return tuples of length 1 or 0 can be handled correctly. This is especially handy, if functions with a variable amount of outputs are wrapped by delayed. E.g. a trivial example: delayed(lambda *args: args, nout=len(vals))(*vals)

Core

• Refactor core byte ingest (GH#1768), (GH#1774)

• Improve import time (GH#1833)

0.12.0 / 2016-11-03

DataFrame

• Return a series when functions given to dataframe.map_partitions return scalars (GH#1515)

• Fix type size inference for series (GH#1513)

• dataframe.DataFrame.categorize no longer includes missing values in the categories. This is for compatibility with a pandas change (GH#1565)

• Fix head parser error in dataframe.read_csv when some lines have quotes (GH#1495)

• Add dataframe.reduction and series.reduction methods to apply generic row-wise reduction to dataframes and series (GH#1483)

• Add dataframe.select_dtypes, which mirrors the pandas method (GH#1556)

• dataframe.read_hdf now supports reading Series (GH#1564)

• Support Pandas 0.19.0 (GH#1540)

• Implement select_dtypes (GH#1556)

• String accessor works with indexes (GH#1561)

• Add pipe method to dask.dataframe (GH#1567)

• Add indicator keyword to merge (GH#1575)

• Support Series in read_hdf (GH#1575)

• Support Categories with missing values (GH#1578)

• Support inplace operators like df.x += 1 (GH#1585)

• Str accessor passes through args and kwargs (GH#1621)

• Improved groupby support for single-machine multiprocessing scheduler (GH#1625)

• Tree reductions (GH#1663)

• Pivot tables (GH#1665)

• Add clip (GH#1667), align (GH#1668), combine_first (GH#1725), and any/all (GH#1724)

• Improved handling of divisions on dask-pandas merges (GH#1666)

• Add groupby.aggregate method (GH#1678)

• Add dd.read_table function (GH#1682)

• Improve support for multi-level columns (GH#1697) (GH#1712)

• Support 2d indexing in loc (GH#1726)

• Extend resample to include DataFrames (GH#1741)

• Support dask.array ufuncs on dask.dataframe objects (GH#1669)

Array

• Add information about how dask.array chunks argument work (GH#1504)

• Fix field access with non-scalar fields in dask.array (GH#1484)

• Add concatenate= keyword to atop to concatenate chunks of contracted dimensions

• Optimized slicing performance (GH#1539) (GH#1731)

• Extend atop with a concatenate= (GH#1609) new_axes= (GH#1612) and adjust_chunks= (GH#1716) keywords

• Add clip (GH#1610) swapaxes (GH#1611) round (GH#1708) repeat

• Automatically align chunks in atop-backed operations (GH#1644)

• Cull dask.arrays on slicing (GH#1709)

Bag

• Fix issue with callables in bag.from_sequence being interpreted as tasks (GH#1491)

• Avoid non-lazy memory use in reductions (GH#1747)

Administration

• Added changelog (GH#1526)

• Create new threadpool when operating from thread (GH#1487)

• Unify example documentation pages into one (GH#1520)

• Add versioneer for git-commit based versions (GH#1569)

• Pass through node_attr and edge_attr keywords in dot visualization (GH#1614)

• Add continuous testing for Windows with Appveyor (GH#1648)

• Remove use of multiprocessing.Manager (GH#1653)

• Add global optimizations keyword to compute (GH#1675)

• Micro-optimize get_dependencies (GH#1722)

0.11.0 / 2016-08-24

Major Points

DataFrames now enforce knowing full metadata (columns, dtypes) everywhere. Previously we would operate in an ambiguous state when functions lost dtype information (such as apply). Now all dataframes always know their dtypes and raise errors asking for information if they are unable to infer (which they usually can). Some internal attributes like _pd and _pd_nonempty have been moved.

The internals of the distributed scheduler have been refactored to transition tasks between explicit states. This improves resilience, reasoning about scheduling, plugin operation, and logging. It also makes the scheduler code easier to understand for newcomers.

Breaking Changes

• The distributed.s3 and distributed.hdfs namespaces are gone. Use protocols in normal methods like read_text('s3://...' instead.

• Dask.array.reshape now errs in some cases where previously it would have create a very large number of tasks

0.10.2 / 2016-07-27

• More Dataframe shuffles now work in distributed settings, ranging from setting-index to hash joins, to sorted joins and groupbys.

• Dask passes the full test suite when run when under in Python’s optimized-OO mode.

• On-disk shuffles were found to produce wrong results in some highly-concurrent situations, especially on Windows. This has been resolved by a fix to the partd library.

• Fixed a growth of open file descriptors that occurred under large data communications

• Support ports in the --bokeh-whitelist option ot dask-scheduler to better routing of web interface messages behind non-trivial network settings

• Some improvements to resilience to worker failure (though other known failures persist)

• You can now start an IPython kernel on any worker for improved debugging and analysis

• Improvements to dask.dataframe.read_hdf, especially when reading from multiple files and docs

0.10.0 / 2016-06-13

Major Changes

• This version drops support for Python 2.6

• Conda packages are built and served from conda-forge

• The dask.distributed executables have been renamed from dfoo to dask-foo. For example dscheduler is renamed to dask-scheduler

• Both Bag and DataFrame include a preliminary distributed shuffle.

Bag

• Add task-based shuffle for distributed groupbys

• Add accumulate for cumulative reductions

DataFrame

• Add a task-based shuffle suitable for distributed joins, groupby-applys, and set_index operations. The single-machine shuffle remains untouched (and much more efficient.)

• Add support for new Pandas rolling API with improved communication performance on distributed systems.

• Add groupby.std/var

• Pass through S3/HDFS storage options in read_csv

• Improve categorical partitioning

• Add eval, info, isnull, notnull for dataframes

Distributed

• Rename executables like dscheduler to dask-scheduler

• Improve scheduler performance in the many-fast-tasks case (important for shuffling)

• Improve work stealing to be aware of expected function run-times and data sizes. The drastically increases the breadth of algorithms that can be efficiently run on the distributed scheduler without significant user expertise.

• Support maximum buffer sizes in streaming queues

• Improve Windows support when using the Bokeh diagnostic web interface

• Support compression of very-large-bytestrings in protocol

• Support clean cancellation of submitted futures in Joblib interface

Other

• All dask-related projects (dask, distributed, s3fs, hdfs, partd) are now building conda packages on conda-forge.

• Change credential handling in s3fs to only pass around delegated credentials if explicitly given secret/key. The default now is to rely on managed environments. This can be changed back by explicitly providing a keyword argument. Anonymous mode must be explicitly declared if desired.

0.9.0 / 2016-05-11

API Changes

• dask.do and dask.value have been renamed to dask.delayed

• dask.bag.from_filenames has been renamed to dask.bag.read_text

• All S3/HDFS data ingest functions like db.from_s3 or distributed.s3.read_csv have been moved into the plain read_text, read_csv functions, which now support protocols, like dd.read_csv('s3://bucket/keys*.csv')

Array

• Add support for scipy.LinearOperator

• Improve optional locking to on-disk data structures

• Change rechunk to expose the intermediate chunks

Bag

• Rename from_filenames to read_text

• Remove from_s3 in favor of read_text('s3://...')

DataFrame

• Fixed numerical stability issue for correlation and covariance

• Allow no-hash from_pandas for speedy round-trips to and from-pandas objects

• Generally reengineered read_csv to be more in line with Pandas behavior

• Support fast set_index operations for sorted columns

Delayed

• Rename do/value to delayed

• Rename to/from_imperative to to/from_delayed

Distributed

• Move s3 and hdfs functionality into the dask repository

• Adaptively oversubscribe workers for very fast tasks

• Improve PyPy support

• Improve work stealing for unbalanced workers

• Scatter data efficiently with tree-scatters

Other

• Add lzma/xz compression support

• Raise a warning when trying to split unsplittable compression types, like gzip or bz2

• Improve hashing for single-machine shuffle operations

• Add new callback method for start state

• General performance tuning

0.8.1 / 2016-03-11

Array

• Bugfix for range slicing that could periodically lead to incorrect results.

• Improved support and resiliency of arg reductions (argmin, argmax, etc.)

Bag

• Add zip function

DataFrame

• Add corr and cov functions

• Add melt function

• Bugfixes for io to bcolz and hdf5

0.8.0 / 2016-02-20

Array

• Changed default array reduction split from 32 to 4

• Linear algebra, tril, triu, LU, inv, cholesky, solve, solve_triangular, eye, lstsq, diag, corrcoef.

Bag

• Add tree reductions

• Add range function

• drop from_hdfs function (better functionality now exists in hdfs3 and distributed projects)

DataFrame

• Refactor dask.dataframe to include a full empty pandas dataframe as metadata. Drop the .columns attribute on Series

• Add Series categorical accessor, series.nunique, drop the .columns attribute for series.

• read_csv fixes (multi-column parse_dates, integer column names, etc. )

• Internal changes to improve graph serialization

Other

• Documentation updates

• Add from_imperative and to_imperative functions for all collections

• Aesthetic changes to profiler plots

• Moved the dask project to a new dask organization

0.7.6 / 2016-01-05

Array

• Improve thread safety

• Tree reductions

• Add view, compress, hstack, dstack, vstack methods

• map_blocks can now remove and add dimensions

DataFrame

• Improve thread safety

• Extend sampling to include replacement options

Imperative

• Removed optimization passes that fused results.

Core

• Removed dask.distributed

• Improved performance of blocked file reading

• Serialization improvements

• Test Python 3.5

0.7.4 / 2015-10-23

This was mostly a bugfix release. Some notable changes:

• Fix minor bugs associated with the release of numpy 1.10 and pandas 0.17

• Fixed a bug with random number generation that would cause repeated blocks due to the birthday paradox

• Use locks in dask.dataframe.read_hdf by default to avoid concurrency issues

• Change dask.get to point to dask.async.get_sync by default

• Allow visualization functions to accept general graphviz graph options like rankdir=’LR’

• Add reshape and ravel to dask.array

• Support the creation of dask.arrays from dask.imperative objects

Deprecation

This release also includes a deprecation warning for dask.distributed, which will be removed in the next version.

Future development in distributed computing for dask is happening here: https://distributed.dask.org . General feedback on that project is most welcome from this community.

0.7.3 / 2015-09-25

Diagnostics

• A utility for profiling memory and cpu usage has been added to the dask.diagnostics module.

DataFrame

This release improves coverage of the pandas API. Among other things it includes nunique, nlargest, quantile. Fixes encoding issues with reading non-ascii csv files. Performance improvements and bug fixes with resample. More flexible read_hdf with globbing. And many more. Various bug fixes in dask.imperative and dask.bag.

0.7.0 / 2015-08-15

DataFrame

This release includes significant bugfixes and alignment with the Pandas API. This has resulted both from use and from recent involvement by Pandas core developers.

• New operations: query, rolling operations, drop

• Improved operations: quantiles, arithmetic on full dataframes, dropna, constructor logic, merge/join, elemwise operations, groupby aggregations

Bag

• Fixed a bug in fold where with a null default argument

Array

• New operations: da.fft module, da.image.imread

Infrastructure

• The array and dataframe collections create graphs with deterministic keys. These tend to be longer (hash strings) but should be consistent between computations. This will be useful for caching in the future.

• All collections (Array, Bag, DataFrame) inherit from common subclass

0.6.1 / 2015-07-23

Distributed

• Improved (though not yet sufficient) resiliency for dask.distributed when workers die

DataFrame

• Improved writing to various formats, including to_hdf, to_castra, and to_csv

• Improved creation of dask DataFrames from dask Arrays and Bags

• Improved support for categoricals and various other methods

Array

• Various bug fixes

• Histogram function

Scheduling

• Added tie-breaking ordering of tasks within parallel workloads to better handle and clear intermediate results

Other

• Added the dask.do function for explicit construction of graphs with normal python code

• Traded pydot for graphviz library for graph printing to support Python3

• There is also a gitter chat room and a stackoverflow tag