duckdb 0.8.1-dev2.0 → 0.8.1-dev219.0

package/src/duckdb/src/execution/window_segment_tree.cpp

@@ -12,7 +12,8 @@ namespace duckdb {
 
 WindowAggregateState::WindowAggregateState(AggregateObject aggr, const LogicalType &result_type_p)
     : aggr(std::move(aggr)), result_type(result_type_p), state(aggr.function.state_size()),
-      statev(Value::POINTER((idx_t)state.data())), statep(Value::POINTER((idx_t)state.data())) {
+      statev(Value::POINTER(CastPointerToValue(state.data()))),
+      statep(Value::POINTER(CastPointerToValue(state.data()))) {
 	statev.SetVectorType(VectorType::FLAT_VECTOR); // Prevent conversion of results to constants
 }
 
@@ -180,11 +181,12 @@ void WindowConstantAggregate::Compute(Vector &target, idx_t rid, idx_t start, id
 //===--------------------------------------------------------------------===//
 // WindowSegmentTree
 //===--------------------------------------------------------------------===//
-WindowSegmentTree::WindowSegmentTree(AggregateObject aggr, const LogicalType &result_type_p, DataChunk *input,
+WindowSegmentTree::WindowSegmentTree(AggregateObject aggr_p, const LogicalType &result_type_p, DataChunk *input,
                                      const ValidityMask &filter_mask_p, WindowAggregationMode mode_p)
-    : aggr(std::move(aggr)), result_type(result_type_p), state(aggr.function.state_size()),
-      statep(Value::POINTER((idx_t)state.data())), frame(0, 0), statev(Value::POINTER((idx_t)state.data())),
-      internal_nodes(0), input_ref(input), filter_mask(filter_mask_p), mode(mode_p) {
+    : aggr(std::move(aggr_p)), result_type(result_type_p), state(aggr.function.state_size()),
+      statep(Value::POINTER(CastPointerToValue(state.data()))), frame(0, 0),
+      statev(Value::POINTER(CastPointerToValue(state.data()))), internal_nodes(0), input_ref(input),
+      filter_mask(filter_mask_p), mode(mode_p) {
 	statep.Flatten(input->size());
 	statev.SetVectorType(VectorType::FLAT_VECTOR); // Prevent conversion of results to constants
 
@@ -212,7 +214,7 @@ WindowSegmentTree::~WindowSegmentTree() {
 	AggregateInputData aggr_input_data(aggr.GetFunctionData(), Allocator::DefaultAllocator());
 	// call the destructor for all the intermediate states
 	data_ptr_t address_data[STANDARD_VECTOR_SIZE];
-	Vector addresses(LogicalType::POINTER, (data_ptr_t)address_data);
+	Vector addresses(LogicalType::POINTER, data_ptr_cast(address_data));
 	idx_t count = 0;
 	for (idx_t i = 0; i < internal_nodes; i++) {
 		address_data[count++] = data_ptr_t(levels_flat_native.get() + i * state.size());

package/src/duckdb/src/function/aggregate/distributive/count.cpp

@@ -7,30 +7,30 @@ namespace duckdb {
 
 struct BaseCountFunction {
 	template <class STATE>
-	static void Initialize(STATE *state) {
-		*state = 0;
+	static void Initialize(STATE &state) {
+		state = 0;
 	}
 
 	template <class STATE, class OP>
-	static void Combine(const STATE &source, STATE *target, AggregateInputData &) {
-		*target += source;
+	static void Combine(const STATE &source, STATE &target, AggregateInputData &) {
+		target += source;
 	}
 
 	template <class T, class STATE>
-	static void Finalize(Vector &result, AggregateInputData &, STATE *state, T *target, ValidityMask &mask, idx_t idx) {
-		target[idx] = *state;
+	static void Finalize(STATE &state, T &target, AggregateFinalizeData &finalize_data) {
+		target = state;
 	}
 };
 
 struct CountStarFunction : public BaseCountFunction {
 	template <class STATE, class OP>
-	static void Operation(STATE *state, AggregateInputData &, idx_t idx) {
-		*state += 1;
+	static void Operation(STATE &state, AggregateInputData &, idx_t idx) {
+		state += 1;
 	}
 
 	template <class STATE, class OP>
-	static void ConstantOperation(STATE *state, AggregateInputData &, idx_t count) {
-		*state += count;
+	static void ConstantOperation(STATE &state, AggregateInputData &, idx_t count) {
+		state += count;
 	}
 
 	template <typename RESULT_TYPE>
@@ -55,27 +55,165 @@ struct CountStarFunction : public BaseCountFunction {
 };
 
 struct CountFunction : public BaseCountFunction {
-	template <class INPUT_TYPE, class STATE, class OP>
-	static void Operation(STATE *state, AggregateInputData &, INPUT_TYPE *input, ValidityMask &mask, idx_t idx) {
-		*state += 1;
+	using STATE = int64_t;
+
+	static void Operation(STATE &state) {
+		state += 1;
 	}
 
-	template <class INPUT_TYPE, class STATE, class OP>
-	static void ConstantOperation(STATE *state, AggregateInputData &, INPUT_TYPE *input, ValidityMask &mask,
-	                              idx_t count) {
-		*state += count;
+	static void ConstantOperation(STATE &state, idx_t count) {
+		state += count;
 	}
 
 	static bool IgnoreNull() {
 		return true;
 	}
+
+	static inline void CountFlatLoop(STATE **__restrict states, ValidityMask &mask, idx_t count) {
+		if (!mask.AllValid()) {
+			idx_t base_idx = 0;
+			auto entry_count = ValidityMask::EntryCount(count);
+			for (idx_t entry_idx = 0; entry_idx < entry_count; entry_idx++) {
+				auto validity_entry = mask.GetValidityEntry(entry_idx);
+				idx_t next = MinValue<idx_t>(base_idx + ValidityMask::BITS_PER_VALUE, count);
+				if (ValidityMask::AllValid(validity_entry)) {
+					// all valid: perform operation
+					for (; base_idx < next; base_idx++) {
+						CountFunction::Operation(*states[base_idx]);
+					}
+				} else if (ValidityMask::NoneValid(validity_entry)) {
+					// nothing valid: skip all
+					base_idx = next;
+					continue;
+				} else {
+					// partially valid: need to check individual elements for validity
+					idx_t start = base_idx;
+					for (; base_idx < next; base_idx++) {
+						if (ValidityMask::RowIsValid(validity_entry, base_idx - start)) {
+							CountFunction::Operation(*states[base_idx]);
+						}
+					}
+				}
+			}
+		} else {
+			for (idx_t i = 0; i < count; i++) {
+				CountFunction::Operation(*states[i]);
+			}
+		}
+	}
+
+	static inline void CountScatterLoop(STATE **__restrict states, const SelectionVector &isel,
+	                                    const SelectionVector &ssel, ValidityMask &mask, idx_t count) {
+		if (!mask.AllValid()) {
+			// potential NULL values
+			for (idx_t i = 0; i < count; i++) {
+				auto idx = isel.get_index(i);
+				auto sidx = ssel.get_index(i);
+				if (mask.RowIsValid(idx)) {
+					CountFunction::Operation(*states[sidx]);
+				}
+			}
+		} else {
+			// quick path: no NULL values
+			for (idx_t i = 0; i < count; i++) {
+				auto sidx = ssel.get_index(i);
+				CountFunction::Operation(*states[sidx]);
+			}
+		}
+	}
+
+	static void CountScatter(Vector inputs[], AggregateInputData &aggr_input_data, idx_t input_count, Vector &states,
+	                         idx_t count) {
+		auto &input = inputs[0];
+		if (input.GetVectorType() == VectorType::FLAT_VECTOR && states.GetVectorType() == VectorType::FLAT_VECTOR) {
+			auto sdata = FlatVector::GetData<STATE *>(states);
+			CountFlatLoop(sdata, FlatVector::Validity(input), count);
+		} else {
+			UnifiedVectorFormat idata, sdata;
+			input.ToUnifiedFormat(count, idata);
+			states.ToUnifiedFormat(count, sdata);
+			CountScatterLoop(reinterpret_cast<STATE **>(sdata.data), *idata.sel, *sdata.sel, idata.validity, count);
+		}
+	}
+
+	static inline void CountFlatUpdateLoop(STATE &result, ValidityMask &mask, idx_t count) {
+		idx_t base_idx = 0;
+		auto entry_count = ValidityMask::EntryCount(count);
+		for (idx_t entry_idx = 0; entry_idx < entry_count; entry_idx++) {
+			auto validity_entry = mask.GetValidityEntry(entry_idx);
+			idx_t next = MinValue<idx_t>(base_idx + ValidityMask::BITS_PER_VALUE, count);
+			if (ValidityMask::AllValid(validity_entry)) {
+				// all valid
+				result += next - base_idx;
+				base_idx = next;
+			} else if (ValidityMask::NoneValid(validity_entry)) {
+				// nothing valid: skip all
+				base_idx = next;
+				continue;
+			} else {
+				// partially valid: need to check individual elements for validity
+				idx_t start = base_idx;
+				for (; base_idx < next; base_idx++) {
+					if (ValidityMask::RowIsValid(validity_entry, base_idx - start)) {
+						result++;
+					}
+				}
+			}
+		}
+	}
+
+	static inline void CountUpdateLoop(STATE &result, ValidityMask &mask, idx_t count,
+	                                   const SelectionVector &sel_vector) {
+		if (mask.AllValid()) {
+			// no NULL values
+			result += count;
+			return;
+		}
+		for (idx_t i = 0; i < count; i++) {
+			auto idx = sel_vector.get_index(i);
+			if (mask.RowIsValid(idx)) {
+				result++;
+			}
+		}
+	}
+
+	static void CountUpdate(Vector inputs[], AggregateInputData &, idx_t input_count, data_ptr_t state_p, idx_t count) {
+		auto &input = inputs[0];
+		auto &result = *reinterpret_cast<STATE *>(state_p);
+		switch (input.GetVectorType()) {
+		case VectorType::CONSTANT_VECTOR: {
+			if (!ConstantVector::IsNull(input)) {
+				// if the constant is not null increment the state
+				result += count;
+			}
+			break;
+		}
+		case VectorType::FLAT_VECTOR: {
+			CountFlatUpdateLoop(result, FlatVector::Validity(input), count);
+			break;
+		}
+		case VectorType::SEQUENCE_VECTOR: {
+			// sequence vectors cannot have NULL values
+			result += count;
+			break;
+		}
+		default: {
+			UnifiedVectorFormat idata;
+			input.ToUnifiedFormat(count, idata);
+			CountUpdateLoop(result, idata.validity, count, *idata.sel);
+			break;
+		}
+		}
+	}
 };
 
 AggregateFunction CountFun::GetFunction() {
-	auto fun = AggregateFunction::UnaryAggregate<int64_t, int64_t, int64_t, CountFunction>(
-	    LogicalType(LogicalTypeId::ANY), LogicalType::BIGINT);
+	AggregateFunction fun({LogicalType(LogicalTypeId::ANY)}, LogicalType::BIGINT, AggregateFunction::StateSize<int64_t>,
+	                      AggregateFunction::StateInitialize<int64_t, CountFunction>, CountFunction::CountScatter,
+	                      AggregateFunction::StateCombine<int64_t, CountFunction>,
+	                      AggregateFunction::StateFinalize<int64_t, int64_t, CountFunction>,
+	                      FunctionNullHandling::SPECIAL_HANDLING, CountFunction::CountUpdate);
 	fun.name = "count";
-	fun.null_handling = FunctionNullHandling::SPECIAL_HANDLING;
 	fun.order_dependent = AggregateOrderDependent::NOT_ORDER_DEPENDENT;
 	return fun;
 }

package/src/duckdb/src/function/aggregate/distributive/first.cpp

@@ -14,9 +14,9 @@ struct FirstState {
 
 struct FirstFunctionBase {
 	template <class STATE>
-	static void Initialize(STATE *state) {
-		state->is_set = false;
-		state->is_null = false;
+	static void Initialize(STATE &state) {
+		state.is_set = false;
+		state.is_null = false;
 	}
 
 	static bool IgnoreNull() {
@@ -27,40 +27,40 @@ struct FirstFunctionBase {
 template <bool LAST, bool SKIP_NULLS>
 struct FirstFunction : public FirstFunctionBase {
 	template <class INPUT_TYPE, class STATE, class OP>
-	static void Operation(STATE *state, AggregateInputData &, INPUT_TYPE *input, ValidityMask &mask, idx_t idx) {
-		if (LAST || !state->is_set) {
-			if (!mask.RowIsValid(idx)) {
+	static void Operation(STATE &state, const INPUT_TYPE &input, AggregateUnaryInput &unary_input) {
+		if (LAST || !state.is_set) {
+			if (!unary_input.RowIsValid()) {
 				if (!SKIP_NULLS) {
-					state->is_set = true;
+					state.is_set = true;
 				}
-				state->is_null = true;
+				state.is_null = true;
 			} else {
-				state->is_set = true;
-				state->is_null = false;
-				state->value = input[idx];
+				state.is_set = true;
+				state.is_null = false;
+				state.value = input;
 			}
 		}
 	}
 
 	template <class INPUT_TYPE, class STATE, class OP>
-	static void ConstantOperation(STATE *state, AggregateInputData &aggr_input_data, INPUT_TYPE *input,
-	                              ValidityMask &mask, idx_t count) {
-		Operation<INPUT_TYPE, STATE, OP>(state, aggr_input_data, input, mask, 0);
+	static void ConstantOperation(STATE &state, const INPUT_TYPE &input, AggregateUnaryInput &unary_input,
+	                              idx_t count) {
+		Operation<INPUT_TYPE, STATE, OP>(state, input, unary_input);
 	}
 
 	template <class STATE, class OP>
-	static void Combine(const STATE &source, STATE *target, AggregateInputData &) {
-		if (!target->is_set) {
-			*target = source;
+	static void Combine(const STATE &source, STATE &target, AggregateInputData &) {
+		if (!target.is_set) {
+			target = source;
 		}
 	}
 
 	template <class T, class STATE>
-	static void Finalize(Vector &result, AggregateInputData &, STATE *state, T *target, ValidityMask &mask, idx_t idx) {
-		if (!state->is_set || state->is_null) {
-			mask.SetInvalid(idx);
+	static void Finalize(STATE &state, T &target, AggregateFinalizeData &finalize_data) {
+		if (!state.is_set || state.is_null) {
+			finalize_data.ReturnNull();
 		} else {
-			target[idx] = state->value;
+			target = state.value;
 		}
 	}
 };
@@ -68,65 +68,64 @@ struct FirstFunction : public FirstFunctionBase {
 template <bool LAST, bool SKIP_NULLS>
 struct FirstFunctionString : public FirstFunctionBase {
 	template <class STATE>
-	static void SetValue(STATE *state, AggregateInputData &input_data, string_t value, bool is_null) {
-		if (LAST && state->is_set) {
-			Destroy(input_data, state);
+	static void SetValue(STATE &state, AggregateInputData &input_data, string_t value, bool is_null) {
+		if (LAST && state.is_set) {
+			Destroy(state, input_data);
 		}
 		if (is_null) {
 			if (!SKIP_NULLS) {
-				state->is_set = true;
-				state->is_null = true;
+				state.is_set = true;
+				state.is_null = true;
 			}
 		} else {
-			state->is_set = true;
-			state->is_null = false;
+			state.is_set = true;
+			state.is_null = false;
 			if (value.IsInlined()) {
-				state->value = value;
+				state.value = value;
 			} else {
 				// non-inlined string, need to allocate space for it
 				auto len = value.GetSize();
 				auto ptr = new char[len];
 				memcpy(ptr, value.GetData(), len);
 
-				state->value = string_t(ptr, len);
+				state.value = string_t(ptr, len);
 			}
 		}
 	}
 
 	template <class INPUT_TYPE, class STATE, class OP>
-	static void Operation(STATE *state, AggregateInputData &input_data, INPUT_TYPE *input, ValidityMask &mask,
-	                      idx_t idx) {
-		if (LAST || !state->is_set) {
-			SetValue(state, input_data, input[idx], !mask.RowIsValid(idx));
+	static void Operation(STATE &state, const INPUT_TYPE &input, AggregateUnaryInput &unary_input) {
+		if (LAST || !state.is_set) {
+			SetValue(state, unary_input.input, input, !unary_input.RowIsValid());
 		}
 	}
 
 	template <class INPUT_TYPE, class STATE, class OP>
-	static void ConstantOperation(STATE *state, AggregateInputData &aggr_input_data, INPUT_TYPE *input,
-	                              ValidityMask &mask, idx_t count) {
-		Operation<INPUT_TYPE, STATE, OP>(state, aggr_input_data, input, mask, 0);
+	static void ConstantOperation(STATE &state, const INPUT_TYPE &input, AggregateUnaryInput &unary_input,
+	                              idx_t count) {
+		Operation<INPUT_TYPE, STATE, OP>(state, input, unary_input);
 	}
 
 	template <class STATE, class OP>
-	static void Combine(const STATE &source, STATE *target, AggregateInputData &input_data) {
-		if (source.is_set && (LAST || !target->is_set)) {
+	static void Combine(const STATE &source, STATE &target, AggregateInputData &input_data) {
+		if (source.is_set && (LAST || !target.is_set)) {
 			SetValue(target, input_data, source.value, source.is_null);
 		}
 	}
 
 	template <class T, class STATE>
-	static void Finalize(Vector &result, AggregateInputData &, STATE *state, T *target, ValidityMask &mask, idx_t idx) {
-		if (!state->is_set || state->is_null) {
-			mask.SetInvalid(idx);
+	static void Finalize(STATE &state, T &target, AggregateFinalizeData &finalize_data) {
+		if (!state.is_set || state.is_null) {
+			finalize_data.ReturnNull();
 		} else {
-			target[idx] = StringVector::AddStringOrBlob(result, state->value);
+			target = StringVector::AddStringOrBlob(finalize_data.result, state.value);
 		}
 	}
 
 	template <class STATE>
-	static void Destroy(AggregateInputData &aggr_input_data, STATE *state) {
-		if (state->is_set && !state->is_null && !state->value.IsInlined()) {
-			delete[] state->value.GetData();
+	static void Destroy(STATE &state, AggregateInputData &aggr_input_data) {
+		if (state.is_set && !state.is_null && !state.value.IsInlined()) {
+			delete[] state.value.GetData();
 		}
 	}
 };
@@ -138,14 +137,14 @@ struct FirstStateVector {
 template <bool LAST, bool SKIP_NULLS>
 struct FirstVectorFunction {
 	template <class STATE>
-	static void Initialize(STATE *state) {
-		state->value = nullptr;
+	static void Initialize(STATE &state) {
+		state.value = nullptr;
 	}
 
 	template <class STATE>
-	static void Destroy(AggregateInputData &aggr_input_data, STATE *state) {
-		if (state->value) {
-			delete state->value;
+	static void Destroy(STATE &state, AggregateInputData &aggr_input_data) {
+		if (state.value) {
+			delete state.value;
 		}
 	}
 	static bool IgnoreNull() {
@@ -153,14 +152,14 @@ struct FirstVectorFunction {
 	}
 
 	template <class STATE>
-	static void SetValue(STATE *state, Vector &input, const idx_t idx) {
-		if (!state->value) {
-			state->value = new Vector(input.GetType());
-			state->value->SetVectorType(VectorType::CONSTANT_VECTOR);
+	static void SetValue(STATE &state, Vector &input, const idx_t idx) {
+		if (!state.value) {
+			state.value = new Vector(input.GetType());
+			state.value->SetVectorType(VectorType::CONSTANT_VECTOR);
 		}
 		sel_t selv = idx;
 		SelectionVector sel(&selv);
-		VectorOperations::Copy(input, *state->value, sel, 1, 0, 0);
+		VectorOperations::Copy(input, *state.value, sel, 1, 0, 0);
 	}
 
 	static void Update(Vector inputs[], AggregateInputData &, idx_t input_count, Vector &state_vector, idx_t count) {
@@ -171,39 +170,32 @@ struct FirstVectorFunction {
 		UnifiedVectorFormat sdata;
 		state_vector.ToUnifiedFormat(count, sdata);
 
-		auto states = (FirstStateVector **)sdata.data;
+		auto states = UnifiedVectorFormat::GetData<FirstStateVector *>(sdata);
 		for (idx_t i = 0; i < count; i++) {
 			const auto idx = idata.sel->get_index(i);
 			if (SKIP_NULLS && !idata.validity.RowIsValid(idx)) {
 				continue;
 			}
-			auto state = states[sdata.sel->get_index(i)];
-			if (LAST || !state->value) {
+			auto &state = *states[sdata.sel->get_index(i)];
+			if (LAST || !state.value) {
 				SetValue(state, input, i);
 			}
 		}
 	}
 
 	template <class STATE, class OP>
-	static void Combine(const STATE &source, STATE *target, AggregateInputData &) {
-		if (source.value && (LAST || !target->value)) {
+	static void Combine(const STATE &source, STATE &target, AggregateInputData &) {
+		if (source.value && (LAST || !target.value)) {
 			SetValue(target, *source.value, 0);
 		}
 	}
 
-	template <class T, class STATE>
-	static void Finalize(Vector &result, AggregateInputData &, STATE *state, T *target, ValidityMask &mask, idx_t idx) {
-		if (!state->value) {
-			// we need to use FlatVector::SetNull here
-			// since for STRUCT columns only setting the validity mask of the struct is incorrect
-			// as for a struct column, we need to also set ALL child columns to NULL
-			if (result.GetVectorType() == VectorType::CONSTANT_VECTOR) {
-				ConstantVector::SetNull(result, true);
-			} else {
-				FlatVector::SetNull(result, idx, true);
-			}
+	template <class STATE>
+	static void Finalize(STATE &state, AggregateFinalizeData &finalize_data) {
+		if (!state.value) {
+			finalize_data.ReturnNull();
 		} else {
-			VectorOperations::Copy(*state->value, result, 1, 0, idx);
+			VectorOperations::Copy(*state.value, finalize_data.result, 1, 0, finalize_data.result_idx);
 		}
 	}
 
@@ -289,7 +281,7 @@ static AggregateFunction GetFirstFunction(const LogicalType &type) {
 		return AggregateFunction({type}, type, AggregateFunction::StateSize<FirstStateVector>,
 		                         AggregateFunction::StateInitialize<FirstStateVector, OP>, OP::Update,
 		                         AggregateFunction::StateCombine<FirstStateVector, OP>,
-		                         AggregateFunction::StateFinalize<FirstStateVector, void, OP>, nullptr, OP::Bind,
+		                         AggregateFunction::StateVoidFinalize<FirstStateVector, OP>, nullptr, OP::Bind,
 		                         AggregateFunction::StateDestroy<FirstStateVector, OP>, nullptr, nullptr);
 	}
 	}

package/src/duckdb/src/function/aggregate/sorted_aggregate_function.cpp

@@ -28,7 +28,7 @@ struct SortedAggregateBindData : public FunctionData {
 		}
 		sorted_on_args = (children.size() == order_bys.orders.size());
 		for (size_t i = 0; sorted_on_args && i < children.size(); ++i) {
-			sorted_on_args = children[i]->Equals(order_bys.orders[i].expression.get());
+			sorted_on_args = children[i]->Equals(*order_bys.orders[i].expression);
 		}
 	}
 
@@ -49,7 +49,7 @@ struct SortedAggregateBindData : public FunctionData {
 	}
 
 	bool Equals(const FunctionData &other_p) const override {
-		auto &other = (const SortedAggregateBindData &)other_p;
+		auto &other = other_p.Cast<SortedAggregateBindData>();
 		if (bind_info && other.bind_info) {
 			if (!bind_info->Equals(*other.bind_info)) {
 				return false;
@@ -253,13 +253,13 @@ struct SortedAggregateState {
 
 struct SortedAggregateFunction {
 	template <typename STATE>
-	static void Initialize(STATE *state) {
-		new (state) STATE();
+	static void Initialize(STATE &state) {
+		new (&state) STATE();
 	}
 
 	template <typename STATE>
-	static void Destroy(AggregateInputData &aggr_input_data, STATE *state) {
-		state->~STATE();
+	static void Destroy(STATE &state, AggregateInputData &aggr_input_data) {
+		state.~STATE();
 	}
 
 	static void ProjectInputs(Vector inputs[], const SortedAggregateBindData &order_bind, idx_t input_count,
@@ -288,7 +288,7 @@ struct SortedAggregateFunction {
 		DataChunk sort_chunk;
 		ProjectInputs(inputs, order_bind, input_count, count, arg_chunk, sort_chunk);
 
-		const auto order_state = (SortedAggregateState *)state;
+		const auto order_state = reinterpret_cast<SortedAggregateState *>(state);
 		order_state->Update(order_bind, sort_chunk, arg_chunk);
 	}
 
@@ -310,7 +310,7 @@ struct SortedAggregateFunction {
 		states.ToUnifiedFormat(count, svdata);
 
 		// Size the selection vector for each state.
-		auto sdata = (SortedAggregateState **)svdata.data;
+		auto sdata = UnifiedVectorFormat::GetDataNoConst<SortedAggregateState *>(svdata);
 		for (idx_t i = 0; i < count; ++i) {
 			auto sidx = svdata.sel->get_index(i);
 			auto order_state = sdata[sidx];
@@ -345,10 +345,10 @@ struct SortedAggregateFunction {
 	}
 
 	template <class STATE, class OP>
-	static void Combine(const STATE &source, STATE *target, AggregateInputData &aggr_input_data) {
-		const auto order_bind = (SortedAggregateBindData *)aggr_input_data.bind_data;
+	static void Combine(const STATE &source, STATE &target, AggregateInputData &aggr_input_data) {
+		auto &order_bind = aggr_input_data.bind_data->Cast<SortedAggregateBindData>();
 		auto &other = const_cast<STATE &>(source);
-		target->Combine(*order_bind, other);
+		target.Combine(order_bind, other);
 	}
 
 	static void Window(Vector inputs[], const ValidityMask &filter_mask, AggregateInputData &aggr_input_data,
@@ -359,7 +359,7 @@ struct SortedAggregateFunction {
 
 	static void Finalize(Vector &states, AggregateInputData &aggr_input_data, Vector &result, idx_t count,
 	                     const idx_t offset) {
-		const auto &order_bind = aggr_input_data.bind_data->Cast<SortedAggregateBindData>();
+		auto &order_bind = aggr_input_data.bind_data->Cast<SortedAggregateBindData>();
 		auto &buffer_manager = order_bind.buffer_manager;
 		RowLayout payload_layout;
 		payload_layout.Initialize(order_bind.arg_types);
@@ -370,7 +370,7 @@ struct SortedAggregateFunction {
 
 		//	 Reusable inner state
 		vector<data_t> agg_state(order_bind.function.state_size());
-		Vector agg_state_vec(Value::POINTER((idx_t)agg_state.data()));
+		Vector agg_state_vec(Value::POINTER(CastPointerToValue(agg_state.data())));
 
 		// State variables
 		auto bind_info = order_bind.bind_info.get();

package/src/duckdb/src/function/cast/cast_function_set.cpp

@@ -158,7 +158,7 @@ int64_t CastFunctionSet::ImplicitCastCost(const LogicalType &source, const Logic
 
 BoundCastInfo MapCastFunction(BindCastInput &input, const LogicalType &source, const LogicalType &target) {
 	D_ASSERT(input.info);
-	auto &map_info = (MapCastInfo &)*input.info;
+	auto &map_info = input.info->Cast<MapCastInfo>();
 	auto entry = map_info.GetEntry(source, target);
 	if (entry) {
 		if (entry->bind_function) {

package/src/duckdb/src/function/cast/decimal_cast.cpp

@@ -207,7 +207,7 @@ struct DecimalCastInput {
 struct StringCastFromDecimalOperator {
 	template <class INPUT_TYPE, class RESULT_TYPE>
 	static RESULT_TYPE Operation(INPUT_TYPE input, ValidityMask &mask, idx_t idx, void *dataptr) {
-		auto data = (DecimalCastInput *)dataptr;
+		auto data = reinterpret_cast<DecimalCastInput *>(dataptr);
 		return StringCastFromDecimal::Operation<INPUT_TYPE>(input, data->width, data->scale, data->result);
 	}
 };

package/src/duckdb/src/function/cast/enum_casts.cpp

@@ -17,7 +17,7 @@ bool EnumEnumCast(Vector &source, Vector &result, idx_t count, CastParameters &p
 	UnifiedVectorFormat vdata;
 	source.ToUnifiedFormat(count, vdata);
 
-	auto source_data = (SRC_TYPE *)vdata.data;
+	auto source_data = UnifiedVectorFormat::GetData<SRC_TYPE>(vdata);
 	auto source_sel = vdata.sel;
 	auto source_mask = vdata.validity;
 
@@ -71,7 +71,7 @@ static bool EnumToVarcharCast(Vector &source, Vector &result, idx_t count, CastP
 	UnifiedVectorFormat vdata;
 	source.ToUnifiedFormat(count, vdata);
 
-	auto source_data = (SRC *)vdata.data;
+	auto source_data = UnifiedVectorFormat::GetData<SRC>(vdata);
 	for (idx_t i = 0; i < count; i++) {
 		auto source_idx = vdata.sel->get_index(i);
 		if (!vdata.validity.RowIsValid(source_idx)) {