duckdb 0.8.2-dev37.0 → 0.8.2-dev3989.0

package/src/duckdb/src/include/duckdb/common/multi_file_reader.hpp

@@ -8,7 +8,7 @@
 
 #pragma once
 
-#include "duckdb/common/types.hpp"
+#include "duckdb/common/common.hpp"
 #include "duckdb/common/multi_file_reader_options.hpp"
 #include "duckdb/common/enums/file_glob_options.hpp"
 #include "duckdb/common/union_by_name.hpp"
@@ -32,6 +32,8 @@ struct HivePartitioningIndex {
 
 	DUCKDB_API void Serialize(Serializer &serializer) const;
 	DUCKDB_API static HivePartitioningIndex Deserialize(Deserializer &source);
+	DUCKDB_API void FormatSerialize(FormatSerializer &serializer) const;
+	DUCKDB_API static HivePartitioningIndex FormatDeserialize(FormatDeserializer &deserializer);
 };
 
 //! The bind data for the multi-file reader, obtained through MultiFileReader::BindReader
@@ -43,6 +45,8 @@ struct MultiFileReaderBindData {
 
 	DUCKDB_API void Serialize(Serializer &serializer) const;
 	DUCKDB_API static MultiFileReaderBindData Deserialize(Deserializer &source);
+	DUCKDB_API void FormatSerialize(FormatSerializer &serializer) const;
+	DUCKDB_API static MultiFileReaderBindData FormatDeserialize(FormatDeserializer &deserializer);
 };
 
 struct MultiFileFilterEntry {
@@ -87,7 +91,8 @@ struct MultiFileReader {
 	DUCKDB_API static vector<string> GetFileList(ClientContext &context, const Value &input, const string &name,
 	                                             FileGlobOptions options = FileGlobOptions::DISALLOW_EMPTY);
 	//! Parse the named parameters of a multi-file reader
-	DUCKDB_API static bool ParseOption(const string &key, const Value &val, MultiFileReaderOptions &options);
+	DUCKDB_API static bool ParseOption(const string &key, const Value &val, MultiFileReaderOptions &options,
+	                                   ClientContext &context);
 	//! Perform complex filter pushdown into the multi-file reader, potentially filtering out files that should be read
 	//! If "true" the first file has been eliminated
 	DUCKDB_API static bool ComplexFilterPushdown(ClientContext &context, vector<string> &files,
@@ -102,7 +107,7 @@ struct MultiFileReader {
 	                                    const MultiFileReaderBindData &options, const string &filename,
 	                                    const vector<string> &local_names, const vector<LogicalType> &global_types,
 	                                    const vector<string> &global_names, const vector<column_t> &global_column_ids,
-	                                    MultiFileReaderData &reader_data);
+	                                    MultiFileReaderData &reader_data, ClientContext &context);
 	//! Create all required mappings from the global types/names to the file-local types/names
 	DUCKDB_API static void CreateMapping(const string &file_name, const vector<LogicalType> &local_types,
 	                                     const vector<string> &local_names, const vector<LogicalType> &global_types,
@@ -157,9 +162,10 @@ struct MultiFileReader {
 	static void InitializeReader(READER_CLASS &reader, const MultiFileReaderOptions &options,
 	                             const MultiFileReaderBindData &bind_data, const vector<LogicalType> &global_types,
 	                             const vector<string> &global_names, const vector<column_t> &global_column_ids,
-	                             optional_ptr<TableFilterSet> table_filters, const string &initial_file) {
+	                             optional_ptr<TableFilterSet> table_filters, const string &initial_file,
+	                             ClientContext &context) {
 		FinalizeBind(options, bind_data, reader.GetFileName(), reader.GetNames(), global_types, global_names,
-		             global_column_ids, reader.reader_data);
+		             global_column_ids, reader.reader_data, context);
 		CreateMapping(reader.GetFileName(), reader.GetTypes(), reader.GetNames(), global_types, global_names,
 		              global_column_ids, table_filters, reader.reader_data, initial_file);
 		reader.reader_data.filters = table_filters;

package/src/duckdb/src/include/duckdb/common/multi_file_reader_options.hpp

@@ -9,8 +9,9 @@
 #pragma once
 
 #include "duckdb/common/types.hpp"
+#include "duckdb/common/case_insensitive_map.hpp"
 #include "duckdb/common/hive_partitioning.hpp"
-#include "re2/re2.h"
+#include "duckdb/main/client_context.hpp"
 
 namespace duckdb {
 class Serializer;
@@ -22,51 +23,20 @@ struct MultiFileReaderOptions {
 	bool hive_partitioning = false;
 	bool auto_detect_hive_partitioning = true;
 	bool union_by_name = false;
+	bool hive_types_autocast = true;
+	case_insensitive_map_t<LogicalType> hive_types_schema;
 
 	DUCKDB_API void Serialize(Serializer &serializer) const;
 	DUCKDB_API static MultiFileReaderOptions Deserialize(Deserializer &source);
+	DUCKDB_API void FormatSerialize(FormatSerializer &serializer) const;
+	DUCKDB_API static MultiFileReaderOptions FormatDeserialize(FormatDeserializer &source);
 	DUCKDB_API void AddBatchInfo(BindInfo &bind_info) const;
-
-	static bool AutoDetectHivePartitioning(const vector<string> &files) {
-		if (files.empty()) {
-			return false;
-		}
-
-		std::unordered_set<string> uset;
-		idx_t splits_size;
-		{
-			//	front file
-			auto splits = StringUtil::Split(files.front(), FileSystem::PathSeparator());
-			splits_size = splits.size();
-			if (splits.size() < 2) {
-				return false;
-			}
-			for (auto it = splits.begin(); it != std::prev(splits.end()); it++) {
-				auto part = StringUtil::Split(*it, "=");
-				if (part.size() == 2) {
-					uset.insert(part.front());
-				}
-			}
-		}
-		if (uset.empty()) {
-			return false;
-		}
-		for (auto &file : files) {
-			auto splits = StringUtil::Split(file, FileSystem::PathSeparator());
-			if (splits.size() != splits_size) {
-				return false;
-			}
-			for (auto it = splits.begin(); it != std::prev(splits.end()); it++) {
-				auto part = StringUtil::Split(*it, "=");
-				if (part.size() == 2) {
-					if (uset.find(part.front()) == uset.end()) {
-						return false;
-					}
-				}
-			}
-		}
-		return true;
-	}
+	DUCKDB_API void AutoDetectHivePartitioning(const vector<string> &files, ClientContext &context);
+	DUCKDB_API static bool AutoDetectHivePartitioningInternal(const vector<string> &files, ClientContext &context);
+	DUCKDB_API void AutoDetectHiveTypesInternal(const string &file, ClientContext &context);
+	DUCKDB_API void VerifyHiveTypesArePartitions(const std::map<string, string> &partitions) const;
+	DUCKDB_API LogicalType GetHiveLogicalType(const string &hive_partition_column) const;
+	DUCKDB_API Value GetHivePartitionValue(const string &base, const string &entry, ClientContext &context) const;
 };
 
 } // namespace duckdb

package/src/duckdb/src/include/duckdb/common/mutex.hpp

@@ -8,6 +8,9 @@
 
 #pragma once
 
+#ifdef __MVS__
+#include <time.h>
+#endif
 #include <mutex>
 
 namespace duckdb {

package/src/duckdb/src/include/duckdb/common/numeric_utils.hpp

@@ -0,0 +1,48 @@
+//===----------------------------------------------------------------------===//
+//                         DuckDB
+//
+// duckdb/common/numeric_utils.hpp
+//
+//
+//===----------------------------------------------------------------------===//
+
+#pragma once
+
+#include <type_traits>
+#include "duckdb/common/hugeint.hpp"
+
+namespace duckdb {
+
+template <class T>
+struct MakeSigned {
+	using type = typename std::make_signed<T>::type;
+};
+
+template <>
+struct MakeSigned<hugeint_t> {
+	using type = hugeint_t;
+};
+
+template <class T>
+struct MakeUnsigned {
+	using type = typename std::make_unsigned<T>::type;
+};
+
+// hugeint_t does not actually have an unsigned variant (yet), but this is required to make compression work
+// if an unsigned variant gets implemented this (probably) can be changed without breaking anything
+template <>
+struct MakeUnsigned<hugeint_t> {
+	using type = hugeint_t;
+};
+
+template <class T>
+struct IsIntegral {
+	static constexpr bool value = std::is_integral<T>::value;
+};
+
+template <>
+struct IsIntegral<hugeint_t> {
+	static constexpr bool value = true;
+};
+
+} // namespace duckdb

package/src/duckdb/src/include/duckdb/common/opener_file_system.hpp

@@ -99,11 +99,15 @@ public:
 	bool IsPipe(const string &filename) override {
 		return GetFileSystem().IsPipe(filename);
 	}
-	virtual void RemoveFile(const string &filename) override {
+	void RemoveFile(const string &filename) override {
 		GetFileSystem().RemoveFile(filename);
 	}
 
-	virtual vector<string> Glob(const string &path, FileOpener *opener = nullptr) override {
+	string PathSeparator(const string &path) override {
+		return GetFileSystem().PathSeparator(path);
+	}
+
+	vector<string> Glob(const string &path, FileOpener *opener = nullptr) override {
 		if (opener) {
 			throw InternalException("OpenerFileSystem cannot take an opener - the opener is pushed automatically");
 		}

package/src/duckdb/src/include/duckdb/common/operator/add.hpp

@@ -11,6 +11,7 @@
 #include "duckdb/common/types.hpp"
 #include "duckdb/common/type_util.hpp"
 #include "duckdb/common/exception.hpp"
+#include "duckdb/common/types/cast_helpers.hpp"
 
 namespace duckdb {
 
@@ -68,14 +69,16 @@ template <>
 bool TryAddOperator::Operation(int32_t left, int32_t right, int32_t &result);
 template <>
 DUCKDB_API bool TryAddOperator::Operation(int64_t left, int64_t right, int64_t &result);
+template <>
+bool TryAddOperator::Operation(hugeint_t left, hugeint_t right, hugeint_t &result);
 
 struct AddOperatorOverflowCheck {
 	template <class TA, class TB, class TR>
 	static inline TR Operation(TA left, TB right) {
 		TR result;
 		if (!TryAddOperator::Operation(left, right, result)) {
-			throw OutOfRangeException("Overflow in addition of %s (%d + %d)!", TypeIdToString(GetTypeId<TA>()), left,
-			                          right);
+			throw OutOfRangeException("Overflow in addition of %s (%s + %s)!", TypeIdToString(GetTypeId<TA>()),
+			                          NumericHelper::ToString(left), NumericHelper::ToString(right));
 		}
 		return result;
 	}

package/src/duckdb/src/include/duckdb/common/operator/cast_operators.hpp

@@ -8,13 +8,18 @@
 
 #pragma once
 
+#include "duckdb/common/assert.hpp"
 #include "duckdb/common/constants.hpp"
+#include "duckdb/common/hugeint.hpp"
 #include "duckdb/common/limits.hpp"
 #include "duckdb/common/exception.hpp"
+#include "duckdb/common/typedefs.hpp"
 #include "duckdb/common/types/string_type.hpp"
 #include "duckdb/common/types.hpp"
 #include "duckdb/common/operator/convert_to_string.hpp"
 #include "duckdb/common/types/null_value.hpp"
+#include "duckdb/common/types/bit.hpp"
+#include "duckdb/common/types/vector.hpp"
 
 namespace duckdb {
 struct ValidityMask;
@@ -68,7 +73,7 @@ struct Cast {
 };
 
 struct HandleCastError {
-	static void AssignError(string error_message, string *error_message_ptr) {
+	static void AssignError(const string &error_message, string *error_message_ptr) {
 		if (!error_message_ptr) {
 			throw ConversionException(error_message);
 		}
@@ -465,6 +470,16 @@ DUCKDB_API bool TryCast::Operation(date_t input, timestamp_t &result, bool stric
 //===--------------------------------------------------------------------===//
 template <>
 DUCKDB_API bool TryCast::Operation(dtime_t input, dtime_t &result, bool strict);
+template <>
+DUCKDB_API bool TryCast::Operation(dtime_t input, dtime_tz_t &result, bool strict);
+
+//===--------------------------------------------------------------------===//
+// Time With Time Zone Casts (Offset)
+//===--------------------------------------------------------------------===//
+template <>
+DUCKDB_API bool TryCast::Operation(dtime_tz_t input, dtime_t &result, bool strict);
+template <>
+DUCKDB_API bool TryCast::Operation(dtime_tz_t input, dtime_tz_t &result, bool strict);
 
 //===--------------------------------------------------------------------===//
 // Timestamp Casts
@@ -474,6 +489,8 @@ DUCKDB_API bool TryCast::Operation(timestamp_t input, date_t &result, bool stric
 template <>
 DUCKDB_API bool TryCast::Operation(timestamp_t input, dtime_t &result, bool strict);
 template <>
+DUCKDB_API bool TryCast::Operation(timestamp_t input, dtime_tz_t &result, bool strict);
+template <>
 DUCKDB_API bool TryCast::Operation(timestamp_t input, timestamp_t &result, bool strict);
 
 //===--------------------------------------------------------------------===//
@@ -501,6 +518,15 @@ DUCKDB_API bool TryCast::Operation(string_t input, dtime_t &result, bool strict)
 template <>
 dtime_t Cast::Operation(string_t input);
 //===--------------------------------------------------------------------===//
+// String -> TimeTZ Casts
+//===--------------------------------------------------------------------===//
+template <>
+DUCKDB_API bool TryCastErrorMessage::Operation(string_t input, dtime_tz_t &result, string *error_message, bool strict);
+template <>
+DUCKDB_API bool TryCast::Operation(string_t input, dtime_tz_t &result, bool strict);
+template <>
+dtime_tz_t Cast::Operation(string_t input);
+//===--------------------------------------------------------------------===//
 // String -> Timestamp Casts
 //===--------------------------------------------------------------------===//
 template <>
@@ -652,6 +678,15 @@ struct CastFromBlob {
 template <>
 duckdb::string_t CastFromBlob::Operation(duckdb::string_t input, Vector &vector);
 
+struct CastFromBlobToBit {
+	template <class SRC>
+	static inline string_t Operation(SRC input, Vector &result) {
+		throw NotImplementedException("Cast from blob could not be performed!");
+	}
+};
+template <>
+string_t CastFromBlobToBit::Operation(string_t input, Vector &result);
+
 struct TryCastToBlob {
 	template <class SRC, class DST>
 	static inline bool Operation(SRC input, DST &result, Vector &result_vector, string *error_message,
@@ -659,7 +694,6 @@ struct TryCastToBlob {
 		throw InternalException("Unsupported type for try cast to blob");
 	}
 };
-
 template <>
 bool TryCastToBlob::Operation(string_t input, string_t &result, Vector &result_vector, string *error_message,
                               bool strict);
@@ -667,14 +701,41 @@ bool TryCastToBlob::Operation(string_t input, string_t &result, Vector &result_v
 //===--------------------------------------------------------------------===//
 // Bits
 //===--------------------------------------------------------------------===//
-struct CastFromBit {
+struct CastFromBitToString {
 	template <class SRC>
 	static inline string_t Operation(SRC input, Vector &result) {
 		throw duckdb::NotImplementedException("Cast from bit could not be performed!");
 	}
 };
 template <>
-duckdb::string_t CastFromBit::Operation(duckdb::string_t input, Vector &vector);
+duckdb::string_t CastFromBitToString::Operation(duckdb::string_t input, Vector &vector);
+
+struct CastFromBitToNumeric {
+	template <class SRC = string_t, class DST>
+	static inline bool Operation(SRC input, DST &result, bool strict = false) {
+		D_ASSERT(input.GetSize() > 1);
+
+		// TODO: Allow conversion if the significant bytes of the bitstring can be cast to the target type
+		// Currently only allows bitstring -> numeric if the full bitstring fits inside the numeric type
+		if (input.GetSize() - 1 > sizeof(DST)) {
+			throw ConversionException("Bitstring doesn't fit inside of %s", GetTypeId<DST>());
+		}
+		Bit::BitToNumeric(input, result);
+		return (true);
+	}
+};
+template <>
+bool CastFromBitToNumeric::Operation(string_t input, bool &result, bool strict);
+template <>
+bool CastFromBitToNumeric::Operation(string_t input, hugeint_t &result, bool strict);
+
+struct CastFromBitToBlob {
+	template <class SRC>
+	static inline string_t Operation(SRC input, Vector &result) {
+		D_ASSERT(input.GetSize() > 1);
+		return StringVector::AddStringOrBlob(result, Bit::BitToBlob(input));
+	}
+};
 
 struct TryCastToBit {
 	template <class SRC, class DST>

package/src/duckdb/src/include/duckdb/common/operator/multiply.hpp

@@ -11,6 +11,7 @@
 #include "duckdb/common/types.hpp"
 #include "duckdb/common/exception.hpp"
 #include "duckdb/common/type_util.hpp"
+#include "duckdb/common/types/cast_helpers.hpp"
 
 namespace duckdb {
 
@@ -64,8 +65,8 @@ struct MultiplyOperatorOverflowCheck {
 	static inline TR Operation(TA left, TB right) {
 		TR result;
 		if (!TryMultiplyOperator::Operation(left, right, result)) {
-			throw OutOfRangeException("Overflow in multiplication of %s (%d * %d)!", TypeIdToString(GetTypeId<TA>()),
-			                          left, right);
+			throw OutOfRangeException("Overflow in multiplication of %s (%s * %s)!", TypeIdToString(GetTypeId<TA>()),
+			                          NumericHelper::ToString(left), NumericHelper::ToString(right));
 		}
 		return result;
 	}

package/src/duckdb/src/include/duckdb/common/operator/numeric_cast.hpp

@@ -9,8 +9,11 @@
 #pragma once
 
 #include "duckdb/common/operator/cast_operators.hpp"
+#include "duckdb/common/types/bit.hpp"
 #include "duckdb/common/types/hugeint.hpp"
+#include "duckdb/common/types/string_type.hpp"
 #include "duckdb/common/types/value.hpp"
+#include "duckdb/common/types/vector.hpp"
 #include <cmath>
 
 namespace duckdb {
@@ -442,6 +445,13 @@ bool TryCastWithOverflowCheck(hugeint_t value, double &result) {
 	return Hugeint::TryCast(value, result);
 }
 
+struct NumericTryCastToBit {
+	template <class SRC>
+	static inline string_t Operation(SRC input, Vector &result) {
+		return StringVector::AddStringOrBlob(result, Bit::NumericToBit(input));
+	}
+};
+
 struct NumericTryCast {
 	template <class SRC, class DST>
 	static inline bool Operation(SRC input, DST &result, bool strict = false) {

package/src/duckdb/src/include/duckdb/common/operator/string_cast.hpp

@@ -71,7 +71,7 @@ struct StringCastTZ {
 template <>
 duckdb::string_t StringCastTZ::Operation(date_t input, Vector &result);
 template <>
-duckdb::string_t StringCastTZ::Operation(dtime_t input, Vector &result);
+duckdb::string_t StringCastTZ::Operation(dtime_tz_t input, Vector &result);
 template <>
 duckdb::string_t StringCastTZ::Operation(timestamp_t input, Vector &result);
 

package/src/duckdb/src/include/duckdb/common/operator/subtract.hpp

@@ -11,6 +11,7 @@
 #include "duckdb/common/types.hpp"
 #include "duckdb/common/exception.hpp"
 #include "duckdb/common/type_util.hpp"
+#include "duckdb/common/types/cast_helpers.hpp"
 
 namespace duckdb {
 
@@ -75,8 +76,8 @@ struct SubtractOperatorOverflowCheck {
 	static inline TR Operation(TA left, TB right) {
 		TR result;
 		if (!TrySubtractOperator::Operation(left, right, result)) {
-			throw OutOfRangeException("Overflow in subtraction of %s (%d - %d)!", TypeIdToString(GetTypeId<TA>()), left,
-			                          right);
+			throw OutOfRangeException("Overflow in subtraction of %s (%s - %s)!", TypeIdToString(GetTypeId<TA>()),
+			                          NumericHelper::ToString(left), NumericHelper::ToString(right));
 		}
 		return result;
 	}

package/src/duckdb/src/include/duckdb/common/optional_idx.hpp

@@ -31,6 +31,9 @@ public:
 	bool IsValid() const {
 		return index != DConstants::INVALID_INDEX;
 	}
+	void Invalidate() {
+		index = INVALID_INDEX;
+	}
 	idx_t GetIndex() {
 		if (index == INVALID_INDEX) {
 			throw InternalException("Attempting to get the index of an optional_idx that is not set");

package/src/duckdb/src/include/duckdb/common/perfect_map_set.hpp

@@ -1,7 +1,7 @@
 //===----------------------------------------------------------------------===//
 //                         DuckDB
 //
-// duckdb/common/string_map_set.hpp
+// duckdb/common/perfect_map_set.hpp
 //
 //
 //===----------------------------------------------------------------------===//
@@ -9,6 +9,7 @@
 #pragma once
 
 #include "duckdb/common/types.hpp"
+#include "duckdb/common/types/validity_mask.hpp"
 #include "duckdb/common/unordered_map.hpp"
 #include "duckdb/common/unordered_set.hpp"
 

package/src/duckdb/src/include/duckdb/common/printer.hpp

@@ -9,6 +9,7 @@
 #pragma once
 
 #include "duckdb/common/common.hpp"
+#include "duckdb/common/string_util.hpp"
 
 namespace duckdb {
 
@@ -21,6 +22,16 @@ public:
 	DUCKDB_API static void Print(OutputStream stream, const string &str);
 	//! Print the object to stderr
 	DUCKDB_API static void Print(const string &str);
+	//! Print the formatted object to the stream
+	template <typename... Args>
+	static void PrintF(OutputStream stream, const string &str, Args... params) {
+		Printer::Print(stream, StringUtil::Format(str, params...));
+	}
+	//! Print the formatted object to stderr
+	template <typename... Args>
+	static void PrintF(const string &str, Args... params) {
+		Printer::PrintF(OutputStream::STREAM_STDERR, str, std::forward<Args>(params)...);
+	}
 	//! Directly prints the string to stdout without a newline
 	DUCKDB_API static void RawPrint(OutputStream stream, const string &str);
 	//! Flush an output stream

package/src/duckdb/src/include/duckdb/common/radix.hpp

@@ -8,6 +8,7 @@
 
 #pragma once
 
+#include "duckdb/common/bswap.hpp"
 #include "duckdb/common/common.hpp"
 #include "duckdb/common/exception.hpp"
 #include "duckdb/common/types.hpp"
@@ -20,18 +21,6 @@
 
 namespace duckdb {
 
-#define BSWAP16(x) ((uint16_t)((((uint16_t)(x)&0xff00) >> 8) | (((uint16_t)(x)&0x00ff) << 8)))
-
-#define BSWAP32(x)                                                                                                     \
-	((uint32_t)((((uint32_t)(x)&0xff000000) >> 24) | (((uint32_t)(x)&0x00ff0000) >> 8) |                               \
-	            (((uint32_t)(x)&0x0000ff00) << 8) | (((uint32_t)(x)&0x000000ff) << 24)))
-
-#define BSWAP64(x)                                                                                                     \
-	((uint64_t)((((uint64_t)(x)&0xff00000000000000ull) >> 56) | (((uint64_t)(x)&0x00ff000000000000ull) >> 40) |        \
-	            (((uint64_t)(x)&0x0000ff0000000000ull) >> 24) | (((uint64_t)(x)&0x000000ff00000000ull) >> 8) |         \
-	            (((uint64_t)(x)&0x00000000ff000000ull) << 8) | (((uint64_t)(x)&0x0000000000ff0000ull) << 24) |         \
-	            (((uint64_t)(x)&0x000000000000ff00ull) << 40) | (((uint64_t)(x)&0x00000000000000ffull) << 56)))
-
 struct Radix {
 public:
 	static inline bool IsLittleEndian() {
@@ -134,19 +123,19 @@ inline void Radix::EncodeData(data_ptr_t dataptr, int8_t value) {
 
 template <>
 inline void Radix::EncodeData(data_ptr_t dataptr, int16_t value) {
-	Store<uint16_t>(BSWAP16(value), dataptr);
+	Store<uint16_t>(BSwap<uint16_t>(value), dataptr);
 	dataptr[0] = FlipSign(dataptr[0]);
 }
 
 template <>
 inline void Radix::EncodeData(data_ptr_t dataptr, int32_t value) {
-	Store<uint32_t>(BSWAP32(value), dataptr);
+	Store<uint32_t>(BSwap<uint32_t>(value), dataptr);
 	dataptr[0] = FlipSign(dataptr[0]);
 }
 
 template <>
 inline void Radix::EncodeData(data_ptr_t dataptr, int64_t value) {
-	Store<uint64_t>(BSWAP64(value), dataptr);
+	Store<uint64_t>(BSwap<uint64_t>(value), dataptr);
 	dataptr[0] = FlipSign(dataptr[0]);
 }
 
@@ -157,17 +146,17 @@ inline void Radix::EncodeData(data_ptr_t dataptr, uint8_t value) {
 
 template <>
 inline void Radix::EncodeData(data_ptr_t dataptr, uint16_t value) {
-	Store<uint16_t>(BSWAP16(value), dataptr);
+	Store<uint16_t>(BSwap<uint16_t>(value), dataptr);
 }
 
 template <>
 inline void Radix::EncodeData(data_ptr_t dataptr, uint32_t value) {
-	Store<uint32_t>(BSWAP32(value), dataptr);
+	Store<uint32_t>(BSwap<uint32_t>(value), dataptr);
 }
 
 template <>
 inline void Radix::EncodeData(data_ptr_t dataptr, uint64_t value) {
-	Store<uint64_t>(BSWAP64(value), dataptr);
+	Store<uint64_t>(BSwap<uint64_t>(value), dataptr);
 }
 
 template <>
@@ -179,13 +168,13 @@ inline void Radix::EncodeData(data_ptr_t dataptr, hugeint_t value) {
 template <>
 inline void Radix::EncodeData(data_ptr_t dataptr, float value) {
 	uint32_t converted_value = EncodeFloat(value);
-	Store<uint32_t>(BSWAP32(converted_value), dataptr);
+	Store<uint32_t>(BSwap<uint32_t>(converted_value), dataptr);
 }
 
 template <>
 inline void Radix::EncodeData(data_ptr_t dataptr, double value) {
 	uint64_t converted_value = EncodeDouble(value);
-	Store<uint64_t>(BSWAP64(converted_value), dataptr);
+	Store<uint64_t>(BSwap<uint64_t>(converted_value), dataptr);
 }
 
 template <>

package/src/duckdb/src/include/duckdb/common/radix_partitioning.hpp

@@ -22,6 +22,9 @@ struct SelectionVector;
 //! Generic radix partitioning functions
 struct RadixPartitioning {
 public:
+	//! 4096 partitions ought to be enough to go out-of-core properly
+	static constexpr const idx_t MAX_RADIX_BITS = 12;
+
 	//! The number of partitions for a given number of radix bits
 	static inline constexpr idx_t NumberOfPartitions(idx_t radix_bits) {
 		return idx_t(1) << radix_bits;
@@ -38,10 +41,12 @@ public:
 		throw InternalException("RadixPartitioning::RadixBits unable to find partition count!");
 	}
 
+	//! Radix bits begin after uint16_t because these bits are used as salt in the aggregate HT
 	static inline constexpr idx_t Shift(idx_t radix_bits) {
-		return 48 - radix_bits;
+		return (sizeof(hash_t) - sizeof(uint16_t)) * 8 - radix_bits;
 	}
 
+	//! Mask of the radix bits of the hash
 	static inline constexpr hash_t Mask(idx_t radix_bits) {
 		return (hash_t(1 << radix_bits) - 1) << Shift(radix_bits);
 	}
@@ -49,26 +54,6 @@ public:
 	//! Select using a cutoff on the radix bits of the hash
 	static idx_t Select(Vector &hashes, const SelectionVector *sel, idx_t count, idx_t radix_bits, idx_t cutoff,
 	                    SelectionVector *true_sel, SelectionVector *false_sel);
-
-	//! Convert hashes to bins
-	static void HashesToBins(Vector &hashes, idx_t radix_bits, Vector &bins, idx_t count);
-};
-
-//! Templated radix partitioning constants, can be templated to the number of radix bits
-template <idx_t radix_bits>
-struct RadixPartitioningConstants {
-public:
-	//! Bitmask of the upper bits of the 5th byte
-	static constexpr const idx_t NUM_PARTITIONS = RadixPartitioning::NumberOfPartitions(radix_bits);
-	static constexpr const idx_t SHIFT = RadixPartitioning::Shift(radix_bits);
-	static constexpr const hash_t MASK = RadixPartitioning::Mask(radix_bits);
-
-public:
-	//! Apply bitmask and right shift to get a number between 0 and NUM_PARTITIONS
-	static inline hash_t ApplyMask(hash_t hash) {
-		D_ASSERT((hash & MASK) >> SHIFT < NUM_PARTITIONS);
-		return (hash & MASK) >> SHIFT;
-	}
 };
 
 //! RadixPartitionedColumnData is a PartitionedColumnData that partitions input based on the radix of a hash

package/src/duckdb/src/include/duckdb/common/row_operations/row_operations.hpp

@@ -13,7 +13,7 @@
 
 namespace duckdb {
 
-class Allocator;
+class ArenaAllocator;
 struct AggregateObject;
 struct AggregateFilterData;
 class DataChunk;
@@ -26,10 +26,10 @@ class Vector;
 struct UnifiedVectorFormat;
 
 struct RowOperationsState {
-	RowOperationsState(Allocator &allocator) : allocator(allocator) {
+	explicit RowOperationsState(ArenaAllocator &allocator) : allocator(allocator) {
 	}
 
-	Allocator &allocator;
+	ArenaAllocator &allocator;
 };
 
 // RowOperations contains a set of operations that operate on data using a RowLayout