v1.17.1

Changes:

• Add in full test matrix
• Docs: Add logging tutorial
• Update warning message for shared memory usage
• Check for _grouping before del

v1.17.0

Changes:

• Docs: dtn.isfinite() and dtn.isnotfinite() rewrite
• Tighten up per-example config setup
• Fix searchsorted for different dtypes
• update col_filter and col_remove
• Docs: Rewrites dtn.resample()
• Add Support for Python-3.12
• Docs: Rewrites dtn.diff()
• Modify isfiltered to handle dict-based categoricals
• Docs: Fixes struct._V validation
• Docs Fixes Struct._A validation
• Docs: Fixes struct._H validation
• Docs: Fixes struct._G validation
• Docs: Add see also link to cat.nunique()
• Pin to riptide_cpp-1.19.0 for Python-3.12

v1.16.1

Changes:

• Changes for NumPy 1.25/1.26
• Replace List[Union[...]] typehints with better alternatives
• Fix rt.Dataset.equals
• Fix bugs with merging 2-dimesional columns
• Accept tuples in accum_cols

v1.16.0

Changes:

• Drop Python 3.9
• Don't set FastArray names when accessing uniquelist
• Docs: DateTimeNano validation (part 3)
• Relax stability requirements on tests
• Docs: Fixes Struct._T validation
• Fix assigning strings to empty dataset
• Docstring validation: Pin the console width/height for docstring validation

v1.15.0

Changes:

• Fix rt.accum_ratiop
• Docs: Fixes validations for methods in Date class
• Testing: treat DeprecationWarning and FutureWarning as errors
• Docs: Struct validation fixes
• Docs: DateTimeNano validation (part 1)
• Docs: DateTimeNano validation (part 2)
• Remove invalid \s from docstring
• New conda index no longer recognizes --no-progress
• Docs: Struct validation (part 2)
• rt_fastarray: allow large_string in from_arrow
• Fix DateTimeNano.info() to show offset correctly.
• Docs: Rewrites DateTimeNano methods (part 1)
• Switch to using conda-index instead of conda-build index
• Docs: rt_sds.save_sds() docstring
• Promote Date value to int to avoid overflow in repr
• Use conda_index module directly
• change some type hints
• Pin to riptide_cpp-1.17.0
• Remove local_scheme for setuptools_scm
• Pin to released rc-1.17.0

v1.14.5

Changes:

• Add tests for empty input and high_card
• Allow rt.Accum2.count to take a filter
• Fix most occurrences of ruff B028.
• Enforce rt.Dataset.load() to return an instance of 'cls'.
• Update all GitHub actions to Node20
• Add more typing annotations to rt.Dataset methods.
• Retry h5 load if multi-threaded load fails

v1.14.4

Changes:

• Remove pandas as implicit pd import
• Docs: Insert the list of valid timezones into docstrings
• Fix the rt_numpy.arange() signature.
• Revert "Fix the rt_numpy.arange() signature."
• Fix arange tests
• Docs: categorical validation firstpass
• Fix pandas import for docstrings
• Allow customizing table HTML output
• Docs: Remove mentions of old codenames from docs and code
• Docs: Updates parameter descriptions for rt.merge
• Don't emit partition with empty pdataset
• Make rt.arange() signature match np.arange()
• Docs : Validation fixes for rt_numpy
• Ignore existing dirs when creating SDS dirs
• Tweaks to build configs
• Enable multi-threaded hdf5 load by default with 16 threads
• Add tests for SDS stacked loading with empty files
• Docs: Fix link from Tutorial/Categoricals to Categoricals User Guide
• Fix checkaddtype for datetime scalars
• Avoid recursing into Datasets in Struct.key_search()
• pin to riptide_cpp-1.16.3
• Fix ruff setting which changed names in recent version of ruff.
• Invoke riptable.tests via pytest, not riptable.tests.run

v1.14.3

Changes:

• Apply groupby sort data to groupby calculation in rt.Accum2
• Fix from_arrow when auto_widen=False for integer arrays
• from_arrow performance
• Don't pretend in the doc string that user provided callables work for .agg
• Fix confusing errors for Dataset construction
• Cat: Remove unused *args and **kwargs
• Fix concat_rows when destroy=True
• fix rt.zeros params
• Support Mumbai TimeZone
• Docs: Move display attributes out of class docstring

v1.14.2

Changes:

• Docs: Fix SDS function docstrings
• Docs: Make "0 or nan" wording consistent in AccumTable
• Have rt.where() handle unsupported calls to rc.Where()
• Remove warning
• Docs: Validate all public functions.
• Fix conda_build-3.28 build break
• Docs: Add repeat() to rt_numpy.py all list so it appears in html
• Docstring: rt_merge.merge2() and page about getting the Cartesian product
• Docs: ds.one_hot_encode()
• Docs: Delete Tutorial sentence about saving/loading multi-key Categoricals
• Add Struct._T/V/H/A/G as public API.
• Simplify Dataset._check_addtype by casting all input types to array
• Docs: Quick fix for Dataset.any
• Docs: Dataset.all quickfix
• Doc Sphinx: Include only public APIs
• Update sphinx requirements
• Fix TimeSpan.from_arrow for null-containing arrays
• Docs: rt_merge.merge_lookup() updates

v1.14.1

Changes:

• [fix] Include zero values in footer dict
• Add python interface for logging
• Fix merge_lookup bug with empty rhs input and add tests
• Add docstring xfail support
• Allow dataset sorts to accept list of booleans for ascending parameter
• Sync up with latest pandas changes
• Docs: Accumtable and functions rewrite
• Docs fastarray validation fix
• Docs fastarraynumba validation fix
• Fix running benchmarks without specifying out option
• Docs: Rewrites dataset.add_matrix
• Pin to pillow=10.0.1=py311h8aef010_1
• Docs: Fix examples for cat.count()
• Docs: Fix examples for cat.as_string_array
• Pin pillow==10.1.0 to avoid solve errors.
• Skip if ipykernel.tests.utils missing
• Eliminate pkg_resources usage
• Docs: Rewrites dataset.all
• Docs: Fix rt_dataset validation errors
• Docs: Rewrites dataset.any
• Pin to riptide_cpp-1.16.1a
• Pin to release riptide_cpp-1.16.1