Changelog

All notable changes to this project will be documented in this file.

Unreleased

• add experimental DuckDB::Result#arrow_c_stream returning DuckDB::ArrowArrayStream to export a query result as an Arrow C stream (Arrow C Data Interface). The stream can be consumed directly by ruby-polars (Polars::DataFrame.new(result)) and red-arrow (Arrow::RecordBatchReader.import(stream.to_i)).

• drop Ruby 3.2.

• add DuckDB::TableFunction::InitInfo#max_threads= (and set_max_threads) to hint DuckDB how many worker threads can execute a custom table function concurrently.

• add DuckDB::TableFunction::InitInfo#column_count to get the number of projected result columns of a custom table function scan.

1.5.3.0 - 2026-05-24

• bump up DuckDB 1.5.3 on CI.

• add DuckDB::AggregateFunctionSet class and DuckDB::Connection#register_aggregate_function_set to register multiple overloads of a custom aggregate function under one SQL name.

• add DuckDB::Appender#append_default_to_chunk.

• add DuckDB::TableNameParser module with shared table name parsing logic (quoting and dot-notation), included by DuckDB::Appender and DuckDB::TableDescription.

• add DuckDB::PreparedStatement#bind_timestamp_tz to bind a TIMESTAMP WITH TIME ZONE (TIMESTAMPTZ) parameter from a Time or timestamp string.

• DuckDB::AggregateFunction#name= and DuckDB::ScalarFunction#name= now accept Symbol arguments (coerced to String).

  Breaking Changes

• DuckDB::ScalarFunction.create: name: is now a required keyword argument (previously optional with nil default). Parameter order changed to name:, return_type:, ....

• DuckDB::ScalarFunctionSet#add: no longer overrides the scalar function’s name with the set’s name. The individual function must have its own name set before being added to the set.

• DuckDB::TableDescription.new: the 2nd argument now parses dot-notation and quoting:

• 'schema.table' is interpreted as schema-qualified (deprecated; use schema: keyword instead).

• '"schema.table"' or "'schema.table'" — quotes are stripped and the name is treated as a literal table name containing a dot.

• DuckDB::Appender.new: the 2nd argument now parses dot-notation and quoting (previously only Connection#appender did this):

• 'schema.table' is interpreted as schema-qualified (deprecated; use schema: keyword instead).

• '"schema.table"' or "'schema.table'" — quotes are stripped and the name is treated as a literal table name containing a dot.

• DuckDB::Connection#appender: table name parsing (dot-notation, quoting) is now delegated to DuckDB::Appender.new internally. Behavior is unchanged — 'schema.table' has always split on dot in Connection#appender. No deprecation warning is emitted.

• DuckDB::Connection#appender: table names surrounded by double or single quotes (e.g. '"a.b"' or "'a.b'") are treated as literal table names — the quotes are stripped and no dot-splitting is performed.

• fix: Connection#appender('a.b') no longer emits a misleading deprecation warning (the behavior was not deprecated — it was already the correct behavior).

• add DuckDB::Appender.new(con, table, schema: nil, catalog: nil) keyword argument form.

• add DuckDB::Connection#appender(table, schema: nil, catalog: nil) keyword argument form.

• deprecate DuckDB::Result#_column_type(i) private method. use columns[i].send(:_type) instead.

• DuckDB::Result#enum_dictionary_values checks invalid column index.

• deprecate DuckDB::Result#_enum_dictionary_size private method.

• deprecate DuckDB::Result#_enum_dictionary_value private method.

  Deprecations

• deprecate DuckDB::Appender.new(con, schema, table) 3-positional-argument form. Use DuckDB::Appender.new(con, table, schema: schema) instead.

• deprecate passing dot-notation string (e.g. 'schema.table') directly to DuckDB::Appender.new. Use DuckDB::Appender.new(con, table, schema: schema) instead.

1.5.2.1 - 2026-04-24

• add DuckDB::AggregateFunction.create.

• add DuckDB::Appender#clear_columns.

• add DuckDB::Appender#add_column.

• add DuckDB::Appender#clear to discard all unflushed data from the appender without writing it to the table (requires DuckDB >= 1.5.0).

• add DuckDB::Value.create_decimal.

• fix gemspec to use git ls-files, preventing precompiled x86_64 .so from being bundled into the gem and breaking non-x86 platforms.

1.5.2.0 - 2026-04-16

• add DuckDB::DataChunk#reset to clear a chunk’s contents so it can be reused across multiple Appender#append_data_chunk calls without reallocation.

• add DuckDB::Database.new(path = :memory, config: nil, &block) to create a database instance. path is a positional argument (default :memory for in-memory database, or a String path for file database). config: accepts an optional DuckDB::Config object. When a block is given, the database is yielded and automatically closed afterwards.

• add TIME_NS column support to DuckDB::Result. TIME_NS values are returned as Time objects with nanoseconds truncated to microseconds.

• add INTERVAL as a supported return_type for DuckDB::ScalarFunction. The scalar function block can now return a DuckDB::Interval object and it will be written back to the result vector correctly.

• add UUID as a supported return_type for DuckDB::ScalarFunction. The scalar function block can now return a UUID string and it will be written back to the result vector correctly.

• add DuckDB::Connection#extract to extract multiple SQL statements as prepared statements without executing them.

• add DuckDB::Connection#register_logical_type to register a custom logical type with a connection.

• add DuckDB::PreparedStatement#bind_value.

• add DuckDB::Value.create_bool.

• add DuckDB::Value.create_int8.

• add DuckDB::Value.create_int16.

• add DuckDB::Value.create_int32.

• add DuckDB::Value.create_int64.

• add DuckDB::Value.create_uint8.

• add DuckDB::Value.create_uint16.

• add DuckDB::Value.create_uint32.

• add DuckDB::Value.create_uint64.

• add DuckDB::Value.create_null.

• add DuckDB::Value.create_float.

• add DuckDB::Value.create_double.

• add DuckDB::Value.create_varchar.

• add DuckDB::Value.create_blob.

• add DuckDB::Value.create_hugeint.

• add DuckDB::Value.create_uhugeint.

• add DuckDB::Appender#append_value.

• add DuckDB::Appender#append_data_chunk to append a DuckDB::DataChunk in one call.

• add DuckDB::DataChunk.new(types) to create an owned data chunk from Ruby logical types.

• add DuckDB::AggregateFunction.

• add DuckDB::TableDescription to retrieve metadata about a table (with DuckDB >= 1.5.0).

• add DuckDB::ColumnDescription to describe a column’s name, logical type, and whether it has a default value (with DuckDB >= 1.5.0).

Breaking Changes

• rename DuckDB::ValueImpl to DuckDB::Value

• remove DuckDB::ExtractedStatementsImpl. use DuckDB::ExtractedStatements

• DuckDB::Database.open to delegate to DuckDB::Database.new. The signature changes from open(dbpath = nil, config = nil) to open(path = :memory, config: nil). DuckDB::Database.open now accepts :memory symbol and config: keyword argument.

1.5.1.1 - 2026-04-04

• fix DuckDB::ScalarFunction to allow HUGEINT and UHUGEINT as return_type and parameter type (the C extension’s vector write path was missing those cases).

• add DuckDB::ScalarFunctionSet to register multiple overloads of a scalar function under one name (wraps duckdb_scalar_function_set).

• add DuckDB::ScalarFunctionSet#add to add a DuckDB::ScalarFunction overload to the set (wraps duckdb_add_scalar_function_to_set).

• add DuckDB::Connection#register_scalar_function_set to register a DuckDB::ScalarFunctionSet with the connection (wraps duckdb_register_scalar_function_set).

• DuckDB::ScalarFunction.create now accepts name: nil (optional) to allow creating nameless functions for use inside a ScalarFunctionSet.

• add DuckDB::LogicalType.create_struct to create a struct logical type.

• add DuckDB::LogicalType.create_enum to create an enum logical type.

• add DuckDB::LogicalType.create_decimal to create a decimal logical type.

1.5.1.0 - 2026-03-29

• add DuckDB::ScalarFunction#varargs_type= to register a scalar function that accepts a variable number of arguments of a given type (wraps duckdb_scalar_function_set_varargs).

• DuckDB::ScalarFunction.create accepts varargs_type: keyword argument (mutually exclusive with parameter_type:/parameter_types:).

• add DuckDB::ScalarFunction#set_special_handling to make the block receive nil for NULL inputs instead of DuckDB short-circuiting (wraps duckdb_scalar_function_set_special_handling).

• DuckDB::ScalarFunction.create accepts null_handling: true keyword to enable special NULL handling.

• add 34 => :any to Converter::IntToSym::HASH_TYPES and :any to ScalarFunction::SUPPORTED_TYPES so DuckDB::LogicalType::ANY can be used with varargs_type=.

• add DuckDB::ScalarFunction#set_bind to register a bind callback (wraps duckdb_scalar_function_set_bind). The block receives a DuckDB::ScalarFunction::BindInfo object at query planning time (once, before per-row execution).

• add DuckDB::ScalarFunction::BindInfo#argument_count to return the number of arguments passed to the scalar function (wraps duckdb_scalar_function_bind_get_argument_count).

• add DuckDB::ScalarFunction::BindInfo#set_error(message) to report an error early at planning time (wraps duckdb_scalar_function_bind_set_error).

• add DuckDB::ScalarFunction::BindInfo#get_argument(index) to return the expression at the given argument index as a DuckDB::Expression object (wraps duckdb_scalar_function_bind_get_argument). Raises ArgumentError for out-of-range index.

• add DuckDB::Expression#foldable? to check whether an expression can be folded to a constant at query planning time (wraps duckdb_expression_is_foldable). Returns true for literals and constant arithmetic, false for column references and non-deterministic functions.

• add DuckDB::LogicalType.create_map to create a map logical type.

• add DuckDB::LogicalType.create_union to create a union logical type.

• bump duckdb to 1.5.1 on CI

• add DuckDB::ScalarFunction::BindInfo#client_context to return the client context of the bind phase as a DuckDB::ClientContext object (wraps duckdb_scalar_function_get_client_context).

• add DuckDB::ClientContext#connection_id to return the connection id of the client context (wraps duckdb_client_context_get_connection_id).

• add DuckDB::Expression#fold(client_context) to fold a constant expression to a native Ruby value at query planning time (wraps duckdb_expression_fold). Returns Integer, Float, String, true/false, or nil. Raises DuckDB::Error if the expression is not foldable.

Breaking changes

• rename DuckDB::BindInfo to DuckDB::TableFunction::BindInfo. DuckDB::BindInfo still works but emits a deprecation warning.

• rename DuckDB::InitInfo to DuckDB::TableFunction::InitInfo. DuckDB::InitInfo still works but emits a deprecation warning.

• rename DuckDB::FunctionInfo to DuckDB::TableFunction::FunctionInfo. DuckDB::FunctionInfo still works but emits a deprecation warning.

1.5.0.2 - 2026-03-22

• add DuckDB.vector_size to return the DuckDB vector size (number of rows processed per vectorized operation).

• DuckDB::ScalarFunction.create, DuckDB::ScalarFunction#add_parameter, DuckDB::ScalarFunction#return_type= accept logical type symbol as argument.

• add DuckDB.default_timezone configuration (defaults to :local) to control how TIMESTAMP and TIME values without time zone are converted to Ruby Time (:utc or :local).

1.5.0.1 - 2026-03-17

• enable DuckDB::ScalarFunction work with duckdb multi threads.

• fix mysql_query fails

1.5.0.0 - 2026-03-15

• bump duckdb to 1.5.0 on CI.

Breaking changes

• remove DuckDB::PrepareadStatement#pending_prepared_stream method. Use DuckDB::PreparedStatement#pending_prepared instead.

• remove DuckDB::Connection#async_query_stream method. Use DuckDB::Connection#async_query instead.

• drop duckdb v1.3.x.

1.4.4.0 - 2026-03-07

• DuckDB::DataChunk#set_value accepts date.

• add DuckDB::LogicalType.create_array to create an array logical type

• add DuckDB::LogicalType.create_list to create a list logical type.

• DuckDB::BindInfo#add_result_column accepts symbols as column, column type argument.

• add DuckDB::LogicalType.resolve.

• add DuckDB.cast.

• support TIMESTAMP_TZ column writing Ruby Time object in DuckDB::DataChunk#set_value.

• add DuckDB::MemoryHelper.write_timestamp_tz method to write a Ruby Time object as a DuckDB TIMESTAMP_TZ (microseconds since Unix epoch UTC) to raw memory.

• support TIMESTAMP column writing Ruby Time object in DuckDB::DataChunk#set_value.

• add DuckDB::MemoryHelper.write_timestamp method to write a Ruby Time object as a DuckDB timestamp to raw memory.

• add DuckDB::Connection#expose_as_table to expose a Ruby object as a queryable DuckDB table function via a registered adapter.

• add DuckDB::TableFunction.add_table_adapter to register a table adapter for a Ruby class.

• add DuckDB::TableFunction.table_adapter_for to look up a registered table adapter by class.

• add DuckDB::TableFunction.create class method for declarative table function creation

• Automatically sets output.size from return value

• Supports positional and named parameters

• Execute block returns row count (0 when done)

• Simplifies table function development with cleaner API

• add DuckDB::DataChunk#set_value method for high-level value writing with automatic type conversion.

• add DuckDB::MemoryHelper write methods for all numeric types (boolean, tinyint, smallint, utinyint, usmallint, uinteger, ubigint, float).

• add DuckDB::Vector#set_validity method to mark rows as NULL or valid.

• add DuckDB::Vector#assign_string_element_len method for BLOB data with null bytes.

• add DuckDB::Vector#logical_type method to get the column type of a vector.

• add DuckDB::TableFunction class (Phase 1: Core container).

• add DuckDB::TableFunction#initialize for standard Ruby allocation pattern.

• add DuckDB::TableFunction#name= for setting function name.

• add DuckDB::TableFunction#add_parameter for positional parameters.

• add DuckDB::TableFunction#add_named_parameter for named parameters.

• add DuckDB::TableFunction#bind for setting bind callback (Phase 2).

• add DuckDB::BindInfo class for table function bind phase (Phase 2).

• add DuckDB::BindInfo#parameter_count for getting parameter count.

• add DuckDB::BindInfo#get_parameter for accessing positional parameters.

• add DuckDB::BindInfo#get_named_parameter for accessing named parameters.

• add DuckDB::BindInfo#add_result_column for defining output schema.

• add DuckDB::BindInfo#set_cardinality for performance hints.

• add DuckDB::BindInfo#set_error for reporting bind errors.

• add DuckDB::DataChunk class for table function output data (Phase 3).

• add DuckDB::DataChunk#column_count for getting number of columns.

• add DuckDB::DataChunk#size for getting number of rows.

• add DuckDB::DataChunk#size= for setting number of rows.

• add DuckDB::DataChunk#get_vector for accessing column vectors.

• add DuckDB::Vector class for column data access (Phase 3).

• add DuckDB::Vector#get_data for raw data pointer access.

• add DuckDB::Vector#get_validity for validity mask access.

• add DuckDB::Vector#assign_string_element for writing string values.

• add DuckDB::MemoryHelper module for writing primitive values to vectors.

• add DuckDB::MemoryHelper.write_bigint for writing BIGINT values.

• add DuckDB::MemoryHelper.write_integer for writing INTEGER values.

• add DuckDB::MemoryHelper.write_double for writing DOUBLE values.

• add DuckDB::TableFunction#execute for setting execute callback (Phase 4).

• add DuckDB::FunctionInfo class for table function execution context (Phase 4).

• add DuckDB::FunctionInfo#set_error for reporting execution errors.

• add DuckDB::TableFunction#init for setting init callback (Phase 5).

• add DuckDB::InitInfo class for table function initialization context (Phase 5).

• add DuckDB::InitInfo#set_error for reporting initialization errors.

• add DuckDB::Connection#register_table_function for registering table functions (Phase 6).

• bump duckdb to 1.4.4 on CI.

• add inline style to DuckDB::Connection#register_scalar_function (accepts keyword arguments + block).

• add DuckDB::ScalarFunction.create class method for declarative API.

• add FLOAT support to DuckDB::ScalarFunction return type.

• add BOOLEAN support to DuckDB::ScalarFunction return type.

• add DOUBLE support to DuckDB::ScalarFunction return type.

• add BIGINT support to DuckDB::ScalarFunction return type.

• refactor DuckDB::ScalarFunction to use vector_set_value_at helper.

• fix DuckDB::ScalarFunction NULL input handling.

• fix DuckDB::ScalarFunction INTEGER output type (int32 vs int64).

• add DuckDB::ScalarFunction#add_parameter.

• add DuckDB::ScalarFunction#set_return_type.

• bump bundler 4.0 in duckdb.gemspec.

1.4.3.0 - 2026-01-10

• bump duckdb to 1.4.3 on CI.

1.4.2.0 - 2025-12-27

• support Ruby 4.0.0, add Ruby 4.0.0 on CI.

• drop duckdb v1.2.x.

• bump duckdb to 1.3.0.

• bump duckdb to 1.4.2 on CI.

1.4.1.1 - 2025-11-03

• DuckDB::Connection#appender_from_query accepts block.

  1.4.1.0 - 2025-11-01

• add DuckDB::Connection#appender_from_query.

• add DuckDB::Appender.create_query.

• add DuckDB::LogicalType.boolean, DuckDB::LogicalType.tinyint, DuckDB::LogicalType.smallint, DuckDB::LogicalType.integer, DuckDB::LogicalType.bigint, DuckDB::LogicalType.utinyint, DuckDB::LogicalType.usmallint, DuckDB::LogicalType.uinteger, DuckDB::LogicalType.ubigint, DuckDB::LogicalType.float, DuckDB::LogicalType.double, DuckDB::LogicalType.timestamp, DuckDB::LogicalType.date, DuckDB::LogicalType.time, DuckDB::LogicalType.interval, DuckDB::LogicalType.hugeint, DuckDB::LogicalType.uhugeint, DuckDB::LogicalType.varchar, DuckDB::LogicalType.blob, DuckDB::LogicalType.timestamp_s, DuckDB::LogicalType.timestamp_ms, DuckDB::LogicalType.timestamp_ns, DuckDB::LogicalType.bit, DuckDB::LogicalType.time_tz, DuckDB::LogicalType.timestamp_tz.

• add DuckDB::LogicalType::BOOLEAN, DuckDB::LogicalType::TINYINT, DuckDB::LogicalType::SMALLINT, DuckDB::LogicalType::INTEGER, DuckDB::LogicalType::BIGINT, DuckDB::LogicalType::UTINYINT, DuckDB::LogicalType::USMALLINT, DuckDB::LogicalType::UINTEGER, DuckDB::LogicalType::UBIGINT, DuckDB::LogicalType::FLOAT, DuckDB::LogicalType::DOUBLE, DuckDB::LogicalType::TIMESTAMP, DuckDB::LogicalType::DATE, DuckDB::LogicalType::TIME, DuckDB::LogicalType::INTERVAL, DuckDB::LogicalType::HUGEINT, DuckDB::LogicalType::UHUGEINT, DuckDB::LogicalType::VARCHAR, DuckDB::LogicalType::BLOB, DuckDB::LogicalType::TIMESTAMP_S, DuckDB::LogicalType::TIMESTAMP_MS, DuckDB::LogicalType::TIMESTAMP_NS, DuckDB::LogicalType::BIT, DuckDB::LogicalType::TIME_TZ, DuckDB::LogicalType::TIMESTAMP_TZ.

• Support TIMESTAMP_NS infinity, -infinity value.

• bump duckdb to 1.3.2, 1.4.1 on CI.

• add DuckDB::ValueImpl class. This class is under construction. You must not use this class directly.

• add DuckDB::ScalarFunction class. This class is under construction.

• add DuckDB::ScalarFunction#set_name.

• add DuckDB::LogicalType.new(type_id) to create a logical type from a type ID.

1.3.1.0 - 2025-06-28

• Support TIMESTAMP_S, TIMESTAMP_MS infinity, -infinity value.

• bump duckdb to 1.3.1 on CI.

• drop dcukdb v1.1.x.

1.3.0.0 - 2025-05-31

• bump duckdb to 1.3.0 on CI.

Breaking changes

• DuckDB::Result#streaming? is deprecated.

• The second argument of DuckDB::PendingResult.new is now meaningless. The result is the same when it is set to true.

• DuckDB::PreparedStatement#pending_prepared behaves the same as DuckDB::PreparedStatement#pending_prepared_stream.

• DuckDB::PreparedStatement#pending_prepared_stream will be depreacted. Use pending_prepared instead.

• DuckDB::Connection#async_query behaves the same as DuckDB::Connection#async_query_stream.

• DuckDB::Connection#async_query_stream will be deprecated. Use async_query instead.

• DuckDB::Result#chunk_each, DuckDB::Result.use_chunk_each, DuckDB::Result.use_chunk_each? are deprecated.

1.2.2.0 - 2025-05-11

• drop Ruby 3.1.

• implement DuckDB::InstanceCache class.

• bump duckdb to 1.2.2 on CI.

• add DuckDB::PreparedStatement#bind_uint8, DuckDB::PreparedStatement#bind_uint16, DuckDB::PreparedStatement#bind_uint32, DuckDB::PreparedStatement#bind_uint64.

• add DuckDB::LogicalType class.

• DuckDB::LogicalType class is under construction. DuckDB::LogicalType#internal_type, DuckDB::LogicalType#dictionary_size, DuckDB::LogicalType#dictionary_value_at, DuckDB::LogicalType#each_dictionary_value, DuckDB::LogicalType#alias, and DuckDB::LogicalType#alias=are available.

1.2.1.0 - 2025-03-30

• bump duckdb v1.2.1 on CI.

• drop duckdb v1.0.0.

• add DuckDB::LogicalType class.

• DuckDB::LogicalType class is under construction. DuckDB::LogicalType#member_count, DuckDB::LogicalType#member_name_at, DuckDB::LogicalType#member_type_at, DuckDB::LogicalType#each_member_name, DuckDB::LogicalType#each_member_type, DuckDB::LogicalType#child_count, DuckDB::LogicalType#child_name_at, DuckDB::LogicalType#child_type_at, DuckDB::LogicalType#each_child_name, and DuckDB::LogicalType#each_child_type are available.

• fix error message when DuckDB::Appender#append_uint16, DuckDB::Appender#append_uint32, DuckDB::Appender#append_uint64, DuckDB::Appender#append_float, DuckDB::Appender#append_double, DuckDB::Appender#append_varchar, DuckDB::Appender#append_varchar_length, DuckDB::Appender#append_blob, DuckDB::Appender#append_null, DuckDB::Appender#append_default, DuckDB::Appender#append_date, DuckDB::Appender#append_interval, DuckDB::Appender#append_time, DuckDB::Appender#append_timestamp, DuckDB::Appender#append_hugeint failed.

• add DuckDB::PreparedStatement#bind_uhugeint.

1.2.0.0 - 2025-02-24

• bump duckdb to 1.2.0.

• add DuckDB::LogicalType class(Thanks to @otegami).

• DuckDB::LogicalType class is under construction. DuckDB::LogicalType#type, DuckDB::LogicalType#width, DuckDB::LogicalType#scale, DuckDB::LogicalType#child_type, DuckDB::LogicalType#size, DuckDB::LogicalType#key_type, and DuckDB::LogicalType#value_type are available.

• add DuckDB::Appender#error_message.

• fix error message when DuckDB::Appender#flush, DuckDB::Appender#close, DuckDB::Appender#end_row, DuckDB::Appender#append_bool, DuckDB::Appender#append_int8, DuckDB::Appender#append_int16, DuckDB::Appender#append_int32, DuckDB::Appender#append_int64, DuckDB::Appender#append_uint8 failed.

• DuckDB::Appender#begin_row does nothing. Only returns self. DuckDB::Appender#end_row is only required.

• bump ruby in CI. use 3.4.2 on MacOS and Ubuntu, 3.4.1 on Windows.

Breaking changes

• DuckDB::Result#row_count, DuckDB::Result#row_size are deprecated.

• DuckDB::Result#use_chunk_each?, DuckDB::Result#use_chunk_each= are deprecated.

• DuckDB::Result#chunk_each is deprecated.

• DuckDB::Result#each only works at first time because duckdb_chunk_each C-API is deprecated. Calling DuckDB::Result#each twice or more does not work. ruby result = con.query('SELECT * FROM table') result.each do |record| p record # <= this works fine. end # calling each again does not work. result.each do |record| p record # <= this will not work end If you prefer to use DuckDB::Result#each multiple times, set DuckDB::Result.use_chunk_each = true. But this behavior will be removed in the future release. ruby DuckDB::Result.use_chunk_each = true result = con.query('SELECT * FROM table') result.each do |record| p record # <= this works fine. end # calling each again works. result.each do |record| p record # <= this works fine. end

• DuckDB::Result#streaming? will be deprecated.

1.1.3.1 - 2024-11-27

• fix to DuckDB::Connection#query with multiple SQL statements. Calling PreparedStatement#destroy after each statement executed.

• install valgrind in docker development environment.

• add DuckDB::Appender#append_default.

1.1.3.0 - 2024-11-10

• add DuckDB::PreparedStatement#bind_decimal. (Thanks to @otegami)

• bump duckdb to 1.1.3.

1.1.2.1 - 2024-11-04

• DuckDB::Connection#query accepts multiple SQL statement.

• When multiple SQL statements are given, DuckDB::Connection#query returns the last statement result.

• DuckDB::Connection#query does not support multiple SQL statements with bind parameters. If you pass 2 or more argument, DuckDB::Connection#query will regard first argument as only one SQL statement and the rest as bind parameters.

• add DuckDB::ExtracteStatements#each method.

• add DuckDB::ExtracteStatementsImpl#destroy method.

• add DuckDB::PreparedStatement#prepare.

• DuckDB::Connection#prepared_statement accepts block and calls PreparedStatement#destroy after block executed.

  con.prepared_statement('SELECT * FROM table WHERE id = ?') do |stmt|
    stmt.bind(1)
    stmt.execute
  end
  

  1.1.2.0 - 2024-10-20

• bump duckdb to 1.1.2.

• add DuckDB::PreparedStatement#destroy.

• DuckDB::Connection#query, DuckDB::Connection#async_query, DuckDB::Connection#async_query_stream call DuckDB::PreparedStatement#destroy to free the prepared statement immediately (#775, #781).

1.1.1.0 - 2024-10-06

• bump duckdb to 1.1.1.

  Breaking changes

• drop duckdb v0.10.x.

1.1.0.1 - 2024-09-21

• add DuckDB::Connection#prepare. DuckDB::Connection#prepare is an alias of DuckDB::Connection#prepared_statement.

1.1.0.0 - 2024-09-15

• drop ruby 3.0.x.

• bump duckdb to 1.1.0.

• Add DuckDB::PendingResult#execute_check_state to check the state of the pending result.

• Add DuckDB::PreparedStatement#clear_bindings to clear all bindings.

• Add DuckDB::PreparedStatement#param_type to get the type of the paramerter type.

• The return value is one of the :invalid, :boolean, :tinyint, :smallint, :integer, :bigint, :utinyint, :usmallint, :uinteger, :ubigint, :float, :double, :timestamp, :date, :time, :interval, :hugeint, :uhugeint, :varchar, :blob, :decimal, :timestamp_s, :timestamp_ms, :timestamp_ns, :enum, :list, :struct, :map, :array, :uuid, :union, :bit, :time_tz, :timestamp_tz.

• Add DuckDB::PreparedStatement#statement_type to get the type of the statement.

• The return value is one of the :invalid, :select, :insert, :update, :explain, :delete, :prepare, :create, :execute, :alter, :transaction, :copy, :analyze, :variable_set, :create_func, :drop, :export, :pragma, :vacuum, :call, :set, :load, :relation, :extension, :logical_plan, :attach, :detach, :multi.

• Add DuckDB::Result#statement_type to get the return type of the statement.

• The return value is one of the :invalid, :select, :insert, :update, :explain, :delete, :prepare, :create, :execute, :alter, :transaction, :copy, :analyze, :variable_set, :create_func, :drop, :export, :pragma, :vacuum, :call, :set, :load, :relation, :extension, :logical_plan, :attach, :detach, :multi.

• Add DuckDB::Result#return_type to get the return type of the result.

• The return value is one of the :invalid, :query_result, :rows_changed, :nothing.

• The following method will be deprecated.

• DuckDB::Result#use_chunk_each?

• DuckDB::Result#use_chunk_each=

Breaking changes

• drop duckdb v0.9.x.

• skip to check duckdb_parameter_name is available.

• The following methods are obsoleted.

• DuckDB::Result#_null?

• DuckDB::Result#_to_boolean

• DuckDB::Result#_to_smallint

• DuckDB::Result#_to_utinyint

• DuckDB::Result#_to_integer

• DuckDB::Result#_to_bigint

• DuckDB::Result#_to_hugeint

• DuckDB::Result#_to_hugeint_internal

• DuckDB::Result#__to_hugeint_internal

• DuckDB::Result#_to_decimal

• DuckDB::Result#_to_decimal_internal

• DuckDB::Result#__to_decimal_internal

• DuckDB::Result#_to_float

• DuckDB::Result#_to_double

• DuckDB::Result#_to_string

• DuckDB::Result#_to_string_internal

• DuckDB::Result#_to_blob

• DuckDB::Result#to_value

• DuckDB::Result#row

• DuckDB::Result#use_chunk_each? is always true.

• DuckDB::Result#use_chunk_each= always ignores the argument.

• DuckDB::Result#each always works the same as DuckDB::Result#chunk_each.

1.0.0.2 - 2024-06-23

• DuckDB::Result supports TIMESTAMPTZ column type (only when DuckDB::Result.use_chunk_each is true).

• Supporting TIMESTAMPTZ is experimental.

• DuckDB::Result supports TIMETZ column type (only when DuckDB::Result.use_chunk_each is true).

• DuckDB::Result supports TIMESTAMP_NS column type (only when DuckDB::Result.use_chunk_each is true).

• DuckDB::Result supports TIMESTAMP_MS column type (only when DuckDB::Result.use_chunk_each is true).

• DuckDB::Result supports TIMESTAMP_S column type (only when DuckDB::Result.use_chunk_each is true).

• DuckDB::Result supports STRUCT column type (only when DuckDB::Result.use_chunk_each is true).

• DuckDB::Result supports MAP column type (only when DuckDB::Result.use_chunk_each is true).

• DuckDB::Result supports UNION column type (only when DuckDB::Result.use_chunk_each is true).

• DuckDB::Result supports BIT column type (only when DuckDB::Result.use_chunk_each is true).

1.0.0.1 - 2024-06-16

• support fetch the value from UHUGEINT type column.

• add DuckDB::Appender#append_uhugeint.

• DuckDB::Result supports ARRAY column type (only when DuckDB::Result.use_chunk_each is true).

• DuckDB::Result supports LIST column type (only when DuckDB::Result.use_chunk_each is true). Thanks to stephenprater.

1.0.0.0 - 2024-06-08

• bump duckdb to 1.0.0.

• add DuckDB::ExtractedStatements class.

• add DuckDB::ExtractedStatements#size.

• add DuckDB::ExtractedStatements#prepared_statement.

• raise error when DuckDB::ExtractedStatements#new is called with invalid SQL.

• The following public/private methods will be deprecated.

• DuckDB::Result#streaming?

• DuckDB::Result#_null?

• DuckDB::Result#_to_boolean

• DuckDB::Result#_to_smallint

• DuckDB::Result#_to_utinyint

• DuckDB::Result#_to_integer

• DuckDB::Result#_to_bigint

• DuckDB::Result#_to_hugeint

• DuckDB::Result#_to_hugeint_internal

• DuckDB::Result#__to_hugeint_internal

• DuckDB::Result#_to_decimal

• DuckDB::Result#_to_decimal_internal

• DuckDB::Result#__to_decimal_internal

• DuckDB::Result#_to_float

• DuckDB::Result#_to_double

• DuckDB::Result#_to_string

• DuckDB::Result#_to_string_internal

• DuckDB::Result#_to_blob

• DuckDB::Result.use_chunk_each=

• DuckDB::Result#use_chunk_each?

Breaking changes

• DuckDB::Result.use_chunk_each is true by default. If you want to use the old behavior, set DuckDB::Result.use_chunk_each = false. But the old behavior will be removed in the future release.

0.10.3.0 - 2024-05-25

• bump to duckdb v0.10.3.

0.10.2.0 - 2024-04-20

• remove version from docker-compose.yml.

• bump to duckdb v0.10.2.

0.10.1.1 - 2024-03-31

• fix error using binding with name (issue #612). Thanks to pere73.

0.10.1.0 - 2024-03-22

• drop duckdb v0.8.x.

• fix column type failures with duckdb v0.10.1.

0.10.0.0 - 2024-02-18

• bump to duckdb v0.10.0.

• fix building error with duckdb v0.10.0.

• bundle update to bump nokogiri from 1.16.0 to 1.16.2.

• fix Decimal type conversion.

Breaking changes

• DuckDB::Connection#query_progress returns DuckDB::QueryProgress object only when duckdb library version is 0.10.0 or later.

• The available methods are DuckDB::QueryProgress#percentage, DuckDB::QueryProgress#rows_processed, DuckDB::QueryProgress#total_rows_to_process.

0.9.2.3 - 2023-12-29

• fix bigdecimal warning with Ruby 3.3.0.

0.9.2.2 - 2023-12-26

• bump Ruby to 3.3.0 on CI.

Breaking changes

• drop Ruby 2.7.

0.9.2.1 - 2023-12-24

• support Time column in DuckDB#Result#chunk_each.

• add DuckDB::Interval#eql?.

0.9.2 - 2023-11-26

• add DuckDB::Connection#async_query_stream.

• DuckDB::PendingResult accepts second argument. If the second argument is true, PendingResult#execute_pending returns streaming DuckDB::Result object.

• add DuckDB::PreparedStatement#pending_prepared_stream

• add DuckDB::Result#streaming?.

0.9.1.2 - 2023-11-05

• add DuckDB::Connection#interrupt, DuckDB::Connection#query_progress.

• add DuckDB::Connection#async_query, alias method async_execute.

0.9.1.1 - 2023-10-29

• change default branch to main from master.

• add DuckDB::PendingResult class.

• add DuckDB::PendingResult#state.

• add DuckDB::PendingResult#execute_task.

• add DuckDB::PendingResult#execute_pending.

• add DuckDB::PreparedStatement#pending_prepared.

Breaking Changes

• drop duckdb v0.7.x.

0.9.1 - 2023-10-14

• add DuckDB::PreparedStatement#parameter_name.

• bump duckdb to 0.9.1.

0.9.0.1 - 2023-10-08

• add DuckDB::PreparedStatement#bind_parameter_index.

• DuckDB::Connection#query accepts SQL with named bind parameters.

0.9.0 - 2023-09-30

• bump duckdb to 0.9.0.

Breaking Changes

• deprecation warning when DuckDB::Result.each calling with DuckDB::Result.use_chunk_each is false. The each behavior will be same as DuckDB::Result.chunk_each in the future. set DuckDB::Result.use_chunk_each = true to suppress the warning.

• DuckDB::Result#chunk_each returns DuckDB::Interval class when the column type is INTERVAL.

0.8.1.3

• Fix BigDecimal conversion.

0.8.1.2

• Fix BigDecimal conversion when the value is 0. Thanks to shreeve.

0.8.1.1

• DuckDB::Result#chunk_each supports:

• UTINYINT

• USMALLINT

• UINTEGER

• UBIGINT

• fix memory leak of:

• DuckDB::Result#_enum_dictionary_value

• DuckDB::Result#_enum_dictionary_size

• DuckDB::Result#_enum_internal_type

0.8.1

• bump duckdb to 0.8.1.

• add DuckDB::Result#chunk_each, DuckDB::Result.use_chunk_each=, DuckDB::Result#use_chunk_each? The current behavior of DuckDB::Result#each is same as older version. But DuckDB::Result#each behavior will be changed like as DuckDB::Result#chunk_each in near future release. And there are some breaking changes. Write DuckdDB::Result.use_chunk_each = true if you want to try new behavior.

  DuckDB::Result.use_chunk_each = true
  
  result = con.query('SELECT ....')
  result.each do |record| # <= each method behavior is same as DuckDB::Result#chunk_each
    ...
  end
  

  Thanks to @stephenprater.

• support enum type in DuckDB::Result#chunk_each.

• support uuid type in DuckDB::Result#chunk_each.

Breaking Changes

• DuckDB::Config.set_config does not raise exception when invalid key specified. Instead, DuckDB::Database.open raises DuckDB::Error with invalid key configuration.

0.8.0

• bump duckdb to 0.8.0

• add DuckDB::Result#_to_decimal_internal

• add DuckDB::Result#_to_hugeint_internal

Breaking Changes

• DuckDB::Result returns BigDecimal object instead of String object if the column type is DECIMAL.

0.7.1

• bump duckdb to 0.7.1

• fix docker build error on M1 Mac

0.7.0

• bump duckdb to 0.7.0

• fix have_func argument order

• remove unused variable in test

• add DuckDB::LIBRARY_VERSION

• add DuckDB::Result#_to_string_internal

• add DuckDB::Result#__to_hugeint_internal

• add DuckDB::Result#__to_decimal_internal

• add Ruby 3.2.1 on CI test

• add Ruby mswin on CI test

  Breaking Changes

• drop Ruby 2.6

0.6.1

• bump Ruby to 3.2.0 on CI

• fix deprected warning (double_heap is deprecated in GC.verify_compaction_references) with Ruby 3.2.0 on CI

• bump duckdb to 0.6.1 on CI

• add DuckDB::PreparedStatement#bind_hugeint_internal

• fix gem install error on M1 MacOS

• implement DuckDB.library_version

• use duckdb_value_string instead of duckdb_value_varchar if duckdb_value_string is available.

• bump Ruby to 3.2.0rc1

• bump duckdb to 0.6.0

Breaking Changes

• drop duckdb <= 0.4.x. ruby-duckdb supports duckdb >= 0.5.0

0.5.1.1

• bug fix: reading the boolean column

0.5.1

• bump duckdb to 0.5.1

0.5.0

• update bundle version of Gemfile.lock

• add ruby ucrt test on Windows

• use TypedData_Wrap_Struct, TypedData_Get_Struct

• bump duckdb to 0.5.0

• fix utf-8 encoding.

• add DuckDB::Result#enum_dictionary_values

• add DuckDB::Result#row_count, DuckDB::Result#row_size(alias of row_count).

• add DuckDB::Result#column_count, DuckDB::Result#column_size(alias of column_count).

Breaking Changes

• bind_varchar does not raised DuckDB::Error when the binding column is date or datetime.

0.3.4.0

• bump duckdb to 0.3.4

0.3.3.0

• DuckDB::Column#type supports :decimal.

• bump duckdb to 0.3.3.

• bump Ruby to 2.6.10, 2.7.6, 3.0.4, 3.1.2.

0.3.2.0

• bind_time, bind_timestamp, bind_date, bind_timeinterval to DuckDB::PreparedStatement

• bump duckdb 0.3.2

• bump Ruby to 3.1.1, add Ruby mingw in CI.

• bump Ruby to 2.6.9, 2.7.5, 3.0.3 in CI.

BREAKING CHANGE

• drop duckdb <= 0.2.8

0.3.1.0

• bump duckdb to 0.3.1 in CI.

• bump duckdb to 0.3.0 in CI.

• append_time, append_date, append_timestamp of DuckDB::Appender accept argument having to_str to convert time string.

0.2.9.0

• add DuckDB::Appender#append

• breaking change.

  • append_timestamp is called when the argument is Time object.

  • append_date is called when the argument is Date object.

• add DuckDB::Appender#append_timestamp.

• add DuckDB::Appender#append_interval. append_interval is expremental.

• add DuckDB::Result#rows_changed

• refactoring DuckDB::Append#append_hugeint with duckdb v0.2.9

• test 2 versions of duckdb on github actions macos CI.

• fix windows CI failes

• update github actions CI on ubuntu

• fix to build with duckdb v0.2.9

• use duckdb_prepare_error when get error message of prepared statement.

• add DuckDB::Appender#append_date

• add DuckDB::Appender#append_time

0.2.8.0

• DuckDB::Database.open accepts 2-nd argument as DuckDB::Config object.

• add DuckDB::Config

• bump duckdb to 0.2.8 in CI

• bump Ruby to 2.6.8, 2.7.4, 3.0.2 in CI

0.2.7.0

• call duckdb_free after calling duckdb_value_blob, duckdb_value_varchar.

• bump DuckDB to v0.2.7 in CI

• rake build on Windows in github actions.

• There is a issue (LoadError) when running rake test on Windows (in GitHub actions).

0.2.6.1

• add DuckDB::PreparedStatement#bind_int8

• DuckDB::Connection#appender accepts block.

• add DuckDB::Appender#append_row.

• support HUGEINT type.

• add DuckDB::Appender#append.

• rename PreparedStatement#bind_boolean to PreparedStatement#bind_bool.

• add DuckDB::Connection#appender.

0.2.6.0

• change version policy

• ruby-duckdb W.X.Y.Z supports duckdb W.X.Y

• add DuckDB::Appender

• bump DuckDB to v0.2.6 in CI.

• remove unnecessary duckdb header file from MacOS CI.

• add DuckDB::Connection#prepared_statement.

0.0.12

• bump DuckDB to v0.2.5

• support BLOB type (with DuckDB version 0.2.5 or later)

0.0.11

• fix failure in test_close in test/duckdb_test/database_test.rb because DuckDB error message was changed.

• bump DuckDb to v0.2.4

• add test CI with Ruby 3.0.0

• add test CI on MacOS.

• bunp DuckDB to v0.2.3

0.0.10

• bump DuckDB to v0.2.2

• fix to build failure on MacOS.

0.0.9

• bump DuckDB to v0.2.1

• bump Ruby to v2.7.2

• bunmp DuckDB to v0.2.0

0.0.8.1

• update Gemfile.lock

• unsupport Ruby 2.4

0.0.8

• remove test with Ruby 2.4.10

• bump DuckDB to v0.1.8

• bump DuckDB to v0.1.8

• bump DuckDB to v0.1.7

• current ruby-duckdb supports DuckDB version 0.1.5 and 0.1.6

• support Ruby 2.7.1

• bump DuckDB to v0.1.6

• current ruby-duckdb supports DuckDB version 0.1.5 and 0.1.6

• DuckDB::Connection#connect accepts block

• add DuckDB::Connection#connect

• DuckDB::Database#connect accepts block

• DuckDB::Database.open accepts block

• update duckdb.gemspec, required ruby version >= 2.4.0

0.0.7

• bump DuckDB to v0.1.5

• DuckDB version must be 0.1.5 or later.

• add DuckDB::Connection#connect, alias method open

• add DuckDB::Connection#disconnect, alias method close

• add DuckDB::Database#close

0.0.6

• add alias execute of DuckDB::Connection#query

• support duckdb version 0.1.3

• add DuckDB:PreparedStatement

• create CI (GitHub Actions / Travis-CI)

• create database only once in result_test.rb

0.0.5

• add DuckDB::Error

• DuckDB::Result#each convert DuckDB number value to Ruby’s number

• DuckDB::Result#each convert DuckDB NULL value to nil

• DuckDB::Result#each returns Enumerator object when block is not given

• DuckDB::Result include Enumerable

• add test for DuckDB::Result

• add test for DuckDB::Connection

• fix description in duckdb.gemspec

0.0.4

• add test for DuckDB::Database

• rename module name to DuckDB from Duckdb

0.0.3

• rename native extension name to duckdb_native

0.0.2

• fix gem install error

0.0.1

• first release