v4.1.0

Changes from v4.0.0:

• Add grouped option for table.objects() method to export nested data. (thanks @lsh!)
• Add shuffle option for the sample() verb. All sampled tables are now shuffled by default.
• Add op.parse_json and op.to_json functions for wrangling JSON string data.
• Add op.map_agg and op.entries_agg aggregate functions, similar to the existing op.object_agg.
• Add columns option for table.objects and table.print output.
• Update table.get(column, row) to default to row index 0 if row is unspecified.
• Update aggregate value collector to avoid unnecessary array copies.
• Fix table expression code generation for computed object keys ({['a'+'b']: 'c'}).
• Fix aq.table() bug with Map input.

v4.0.0

Changes from v3.0.0:

• (Breaking Change) Add op.array_agg aggregation function, remove prior op.values definition .
• (Breaking Change) Add op.array_agg_distinct aggregation function, remove prior op.unique definition.
• Add op.object_agg aggregate function.
• Add op.greatest and op.least functions, corresponding to Math.max and Math.min.
• Add op.compact function to filter invalid values from an array.
• Add support for Map and Set instances to op.has(), op.keys(), and op.values().
• Add op.entries() function, with support for Object, Map, and Set instances.
• Add table columnArray() method.
• Update parser to rewrite Math.max and Math.min to use greatest and least.
• Fix dedupe() to produce ungrouped tables.

v3.0.0

Summary:

• (Breaking Change) Update table getter methods to respect filter and orderby.
• Add the arquero-arrow package into the core Arquero package.
• Add file loading methods for URLs and local files.
• Update build system to use separate node.js and browser builds.

Changes from v2.3.0:

• Breaking: Update table get() and getter() to respect filter and orderby settings.
• Add load(), loadArrow(), loadCSV(), and loadJSON() file loading methods.
• Add table toArrow() method.
• Add table toArrowBuffer() method.
• Add top-level toArrow() method that encodes either Arquero tables or arrays-of-objects.
• Add top-level Type object to export Arrow type ids.
• Add table size property, which matches numRows(). (thanks @mbostock!)
• Update fromArrow() to also support Arrow buffer byte array input.
• Update fromJSON() for consistent auto-typing and to improve performance.
• Fix table reify() bug for Arrow columns with null values. (See also #99)
• Split node.js and browser builds.
• Add tests.

v2.3.0

Changes from v2.2.0:

• Add an internal dictionary column abstraction.
• Add "pay-as-you-go" extractors for nested Arrow types, rather than unpacking.
• Add inlined table scans in performance critical sections (includes derive, filter, groupby, and aggregate).
• Add extensibility API for custom table methods, verbs, and package bundles.
• Update expression parser to optimize equality checks for dictionary/literal comparisons.
• Update expression parser to annotate direct column and operator references.
• Update groupby() verb to directly use dictionary keys when applicable.
• Update rollup() logic to generate table-compatible column arrays and assign them when applicable.
• Update from() with a more performant path for ingesting an array of objects.
• Add performance benchmarks for direct Arrow and Arrow-Arquero operations.
• Add performance benchmarks for table-object serialization.
• Fix AST tree walker bug for list types.

v2.2.0

Changes from v2.1.3:

• Update table() constructor to accept Map objects and [[key, value], ...] entry arrays as input.
• Update fromCSV() parsing to ensure consistent types across column values.
• Add autoMax option to fromCSV() to control how many initial rows are used for type inference.
• Add assign() method to bind columns from multiple tables into a new table.
• Add slice() verb to extract contiguous rows from table groups.
• Add frac() helper to generate expressions for per-group row counts. Useful as input to the sample() verb.
• Add column iterator support. (thanks @mbostock!)
• Add column factory mechanism, export as internal.columnFactory() getter/setter.
• Add exports for internal.Transformable and internal.Verb classes.
• Add internal performance benchmark scripts.
• Refactor: Consolidate default null / missing value.
• Refactor: Consolidate internal loop unrolling utilities.
• Refactor: Reorganize directory structure.

v2.1.3

Changes from v2.1.2:

• Fix expression parser to evaluate non-literal column reference expressions. (#74)

v2.1.2

Changes from v2.1.1:

• Update JSDoc comments and improve generated TypeScript typings.
• Update Table API docs.
• Fix table iterator filter check bug.

v2.1.1

Changes from v2.1.0:

• Fix JSDoc comments for table() method. (thanks @vsimko!)

v2.1.0

Changes from v2.0.0:

• Add index argument to op.match().
• Fix table indices() to properly handle order argument.
• Fix fromJSON() auto-type parsing of integer strings.
• Fix fromArrow() handling of FixedSizeList columns.

v2.0.0

Changes from v1.3.1:

• Breaking: Add schema metadata to fromJSON() and toJSON().
• Breaking: Add drop option to spread() verb to drop input columns, set default to true.
• Breaking: Revise spread() to use undefined for missing values.
• Breaking: Revise spread() to use _ separator between column name and index number in auto-generated names.
• Breaking: Revise spread() to use the length of the as option as an implicit limit when limit is unspecified.
• Breaking: Revise pivot() to place value name before key name when multiple values are specified.
• Breaking: Consolidate internal Query and QueryBuilder classes.
• Add impute() verb.
• Add column reference checks to pivot() and rollup().
• Add more flexible join key specification for lookup().
• Update sample() weight calculation to drop unneeded columns.
• Update verb JSDoc comments / TypeScript types.
• Fix input type checks for sequence op functions.
• Fix code generation for safer output.
• Refactor: Expose row_number op for internal use.
• Refactor: Update expression parser error messages.
• Refactor: Add Transformable base class for both Table and Query.
• Refactor: Update ColumnTable constructor.