v4.0.3

Open source release 4.0.3

• Changed info string to standard representation for uninitialized DataFrames in info() method.
• Fixed some errors in Column conversions.

v4.0.2

Open source release 4.0.2

• Changed NullableDataFrame.addColumn() methods to cache row objects
• Changed AbstractDataFrame.toArray() to avoid unnecessary clone operation
• Changed exception error messages in DataFrame implementations

v4.0.1

Open source release 4.0.1

• Added checks for unsigned ints for row and column count in DataFrame serialization
• Added additional catch clauses in DataFrameSerializer.serialize() and DataFrameSerializer.deserialize() to catch and re-throw all SerializationExceptions
• Added null checks in DataFrameUtils.merge() function for every argument array element
• Changed DataFrameSerializer.serializeImplv2() to use for-loop instead of for-each-loop over DataFrame columns to avoid additional copy operations
• Changed DataFrameSerializer.copyBytes() to use native System.arraycopy() for improved performance
• Fixed a benign bug in DataFrame serialization which in some cases added a superfluous zero byte to the end of the array
• Fixed a bug in NullableDataFrame minimum() and maximum() methods which returned NaN when there is only one non-null max/min value
• Fixed column handling when adding, inserting and setting empty columns
• Fixed return value of NullableDataFrame.castToNumericType() to null for integer cases when double value is NaN
• Fixed incorrect NaN presorting in float and double columns

v4.0.0

Open source release 4.0.0

• Changed DataFrame API and implementations to strictly adhere to official specification
• Fixed a bug in reading of large DataFrame files
• Improved reading of CSV files with malformed formats
• Removed DataFrame serialization support for the legacy text-based version 1 format

v3.0.1

Open source release 3.0.1

• Added BitVector implementation
• Added ProbabilisticSet interface and implementing BloomFilter classes
• Added Item, WritableItem, FinalItem and ObservableItem classes
• Added Bit class with static constants
• Added Serializer interface
• Added StringSerializer class
• Added DataFrameSerializerInstance class
• Added SerializationException class
• Added ArgumentParser optional methods to remove boolean trap
• Added Actable, Action and TimedAction interfaces
• Added FutureAction class
• Added Chronometer support for FutureActions
• Added ByteFunction interface
• Added getType() method in abstract Column class
• Added BinaryColumn and NullableBinaryColumn classes to DataFrame API
• Added withHeader() method to CSVReader and CSVWriter class and remove boolean trap from constructor
• Added DataFrameSerializer and CSVReader CSVWriter read and write support from input/outputStreams
• Added ConfigurationFile complete review
• Added ConfigurationFile and PropertiesFile read and write support from input/outputStreams
• Added Hash class with static utility methods
• Added HashFunction enum
• Changed DataFrame addRow() and such method argument to varargs
• Changed indexOfAll() method in DataFrame API to return an empty array if nothing found
• Changed clone() return type to Column in abstract Column class
• Changed ConfigurationFile.Section class to be iterable
• Changed return type to CompletableFuture for async methods in CSVReader, CSVWriter and DataFrameSerializer
• Fixed Chronometer formatting bug for elapsed hours
• Fixed bug in equals() method in DefaultDataFrame and NullableDataFrame implementation
• Refactored static DateFrame utility methods to DataFrameUtil class
• Removed ConcurrentReader and ConcurrentWriter interfaces

v2.5.0

Open source release 2.5.0

• Added serialization support for the DataFrame version 2 binary format (v2)
• Added unique type codes to all classes extending Column. This is now the preferred way for differentiating columns instead of using the instanceof operator
• Added some JUnit tests for the made changes
• Changed namespace from com.kilo52.common to com.raven.common
• Changed DataFrameSerializer so that all operations are now provided as public static methods. The class cannot be instantiated anymore
• Changed the DataFrame API so that references of column labels are now located directly in each Column instance. Appropriate constructors and methods were added
• Changed CSVFileReader and CSVFileWriter so that headers and data values are properly escaped when they contain instances of the used separator
• Removed the Serializable interface extension from the DataFrame interface and Column abstract class. Serialization is now officially only supported via the DataFrameSerializer class