v0.4.10

This release achieves compatibility with the recent versions of fsspec and pandas. Thanks to the new contributor @philipc2, and also @maximlt and @ianthomas23.

Enhancements:

• Construct PolygonArray or MultiPolygonArray from Exterior Coordinates (#312)
• Improvements to CI (#129 and #134)

Compatibility:

• Adapt glob to be compatible with the latest fsspec version (#133)
• Drop support for Python 3.8 (#131)
• Remove some pandas 2.1 future warnings (#130)

Full Changelog: v0.4.9...v0.4.10

Version 0.4.9

This is a compatibility release to support Pandas 2.1.

Compatibility:

• Support Pandas 2.1 (#125)
• Don't compare empty geometry tests against geopandas (#127)

Enhancements:

• Rename pyviz-dev as holoviz-dev (#120)
• Use holoviz_tasks/install action for CI (#123)

Thanks to @maximlt, @hoxbro and @ianthomas23.

Version 0.4.8

This is a compatibility release to support numba 0.57 and Python 3.11, and fixes the reading of parquet files from remote filesystems.

Compatibility:

• Check compatibility with numba 0.57 and python 3.11 (#113)
• Fix read parquet from remote filesystems (#115)
• Set python-snappy as optional dependency (#117)
• Remove pathlib.Path.enter deprecation warnings (#118)

Enhancements:

• Improvements to CI (#112 and #114)

Thanks to new contributor @weiji14 and also @brl0 @philippjfr, and @ianthomas23

Version 0.4.7

This is a compatibility release to support numpy 1.24 and pyarrow from version 1.0.0 to 11.0.0.

Compatibility:

• Drop support for pyarrow<0.16 (#108)
• Remove all use of parquet's validate_schema (#110)
• Support pyarrow back to version 1.0.0 (#111)

Thanks to new contributor @gsakkis and also @ianthomas23.

Version 0.4.6

This is a compatibility release to provide support for numpy 1.24.

Compatibility:

• Support numpy 1.24 ragged array conversion (#107)

Enhancements:

• Rename github default branch from master to main (#105)

Thanks to @ianthomas23 and @maximlt.

Version 0.4.5

This is a compatibility release to provide support for Shapely 2.0.

Compatibility:

• Don't use dtype=object when converting arrays to pass to shapely 2 (#103)

Enhancements:

• Improvements to CI (#97, #98, #99, #100)

Thanks to @ianthomas23 and @maximlt.

Version 0.4.4

This release achieves compatibility with recent versions of Dask, Pandas, PyArrow and Shapely.

Enhancements:

• Avoid inappropriate geometry tests (#90)

Compatibility:

• Fix Shapely deprecation warnings (#85)
• Pandas extension array fixes (#88)
• PyArrow and Dask parquet issues (#92)

Version 0.4.3

Enhancements:

• Expands the optional arguments that can be passed to to_parquet, to_parquet_dask, read_parquet, read_parquet_dask ensuring that storage_options is appropriately passed where needed. (#79)
• Add typing information to some functions.
• Update build_sindex to pass through kwargs to underlying HilbertRtree implementation.
• Change build_sindex methods to consistently return reference to object to allow for method chaining.

Bug fixes:

• Update internal validate_coerce_filesystem to pass storage_options through. (#78)

Compatibility:

• Adds the ability to pass storage_options to the to_parquet function for pandas > 1.2, otherwise instantiates an fsspec filesystem with storage_options and passes that.
• Renames fname parameter to path to align with the pandas convention.

Version 0.4.3a1

v0.4.3a1

Update CHANGELOG (#82)

Version 0.4.2

This release primarily achieves compatibility with recent releases of Pandas. Many thanks to @hoxbro for contributing the fixes and @philippjfr for ongoing maintenance of the project.

Compatibility:

• Compatibility with Pandas>=1.3 (#76)