v0.32.1

• Introduce ReaderRegistry and new DataFrameReader methods

v0.32.0 - XLSX support and optional modules (broken)

Major features

• XLSX support (#470). Thanks @hallvard
• New optional module structure, which was unfortunately broken (#475)

Enhancements

• Implemented plot.ly's categoryOrder
• Support comment character in CSV files (#483). Thanks @jmcgonegal
• Plot.ly template customization (#482). Thanks @hallvard
• Configure shade plugin to create OSGi bundles (#481). Thanks @hallvard

Bug fixes

• Fixed stepWithRows

Development

• Upgrade to junit 5 (#476)

v0.31.0

• Fat columns returned from file readers by default
• Fixed-width file support (Thanks @Sparow199)
• Improved marker support
• Support for custom color bars on scatter charts with scaled colors
• Additional table creation validation
• Fixed numeric to string conversion

v0.30.4

• Adds ability to read and write JSON data

v0.30.3

• Fix for NullPointerException in Table.read().csv()
• Fix for marker color
• Minor join performance improvements
• Fix for Windows line ending issues
• Fix for reading SQL result set
• Allow pie charts to take floating points
• Fixed sorting in TimeSeriesPlot

v0.30.2

• Improved Smile integration
• Pivot table functionality
• Bug fix for conversion from Hive bigInt
• SQL number type mapping. Thanks @gregorco
• Bug fix for unique()
• Improved null handling for StringColumn
• SQL Date support

v0.30.1

• Tweaks to CSV sampling
• Fix isMissing for StringColumn and TextColumn. Thanks @armarti

v0.30.0: Major update to column API

With this release we wanted to strike a balance between making the implementation and maintenance of the library more approachable, reducing memory usage, and providing powerful operations.

We provide non-double numeric column types so that data can be held in memory in a less memory-intensive manner. All computations return DoubleColumn. The alternative would be supporting every pair-wise type of numeric, which would quickly make the library too difficult to contribute to and maintain. The user can convert DoubleColumn to less memory intensive column types like IntColumn as they choose to reduce in-memory overhead for large tables. For small tables, users may be comfortable keeping all columns as DoubleColumn.

Other improvements include:

• Improvements to Smile machine learning integration
• Additional CSV parsing options
• Bug fixes

Thanks to @hallvard, @Ebalaitung, @jln-ho, @gregorco for contributing to this release

v0.25.2

• Fixed generics warnings
• CSV read performance enhancement

v0.25.1

• Switched to Univocity library for CSV handling