duckdb 0.7.2-dev225.0 → 0.7.2-dev2320.0

package/src/duckdb/src/optimizer/statistics/expression/propagate_comparison.cpp

@@ -1,15 +1,14 @@
 #include "duckdb/optimizer/statistics_propagator.hpp"
 #include "duckdb/planner/expression/bound_comparison_expression.hpp"
 #include "duckdb/planner/expression/bound_constant_expression.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
 #include "duckdb/optimizer/expression_rewriter.hpp"
 
 namespace duckdb {
 
-FilterPropagateResult StatisticsPropagator::PropagateComparison(BaseStatistics &left, BaseStatistics &right,
+FilterPropagateResult StatisticsPropagator::PropagateComparison(BaseStatistics &lstats, BaseStatistics &rstats,
                                                                 ExpressionType comparison) {
 	// only handle numerics for now
-	switch (left.type.InternalType()) {
+	switch (lstats.GetType().InternalType()) {
 	case PhysicalType::BOOL:
 	case PhysicalType::UINT8:
 	case PhysicalType::UINT16:
@@ -26,9 +25,7 @@ FilterPropagateResult StatisticsPropagator::PropagateComparison(BaseStatistics &
 	default:
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	}
-	auto &lstats = (NumericStatistics &)left;
-	auto &rstats = (NumericStatistics &)right;
-	if (lstats.min.IsNull() || lstats.max.IsNull() || rstats.min.IsNull() || rstats.max.IsNull()) {
+	if (!NumericStats::HasMinMax(lstats) || !NumericStats::HasMinMax(rstats)) {
 		// no stats available: nothing to prune
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	}
@@ -38,52 +35,53 @@ FilterPropagateResult StatisticsPropagator::PropagateComparison(BaseStatistics &
 	switch (comparison) {
 	case ExpressionType::COMPARE_EQUAL:
 		// l = r, if l.min > r.max or r.min > l.max equality is not possible
-		if (lstats.min > rstats.max || rstats.min > lstats.max) {
+		if (NumericStats::Min(lstats) > NumericStats::Max(rstats) ||
+		    NumericStats::Min(rstats) > NumericStats::Max(lstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		} else {
 			return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 		}
 	case ExpressionType::COMPARE_GREATERTHAN:
 		// l > r
-		if (lstats.min > rstats.max) {
+		if (NumericStats::Min(lstats) > NumericStats::Max(rstats)) {
 			// if l.min > r.max, it is always true ONLY if neither side contains nulls
 			return has_null ? FilterPropagateResult::FILTER_TRUE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_TRUE;
 		}
 		// if r.min is bigger or equal to l.max, the filter is always false
-		if (rstats.min >= lstats.max) {
+		if (NumericStats::Min(rstats) >= NumericStats::Max(lstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		}
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	case ExpressionType::COMPARE_GREATERTHANOREQUALTO:
 		// l >= r
-		if (lstats.min >= rstats.max) {
+		if (NumericStats::Min(lstats) >= NumericStats::Max(rstats)) {
 			// if l.min >= r.max, it is always true ONLY if neither side contains nulls
 			return has_null ? FilterPropagateResult::FILTER_TRUE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_TRUE;
 		}
 		// if r.min > l.max, the filter is always false
-		if (rstats.min > lstats.max) {
+		if (NumericStats::Min(rstats) > NumericStats::Max(lstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		}
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	case ExpressionType::COMPARE_LESSTHAN:
 		// l < r
-		if (lstats.max < rstats.min) {
+		if (NumericStats::Max(lstats) < NumericStats::Min(rstats)) {
 			// if l.max < r.min, it is always true ONLY if neither side contains nulls
 			return has_null ? FilterPropagateResult::FILTER_TRUE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_TRUE;
 		}
 		// if l.min >= rstats.max, the filter is always false
-		if (lstats.min >= rstats.max) {
+		if (NumericStats::Min(lstats) >= NumericStats::Max(rstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		}
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
 	case ExpressionType::COMPARE_LESSTHANOREQUALTO:
 		// l <= r
-		if (lstats.max <= rstats.min) {
+		if (NumericStats::Max(lstats) <= NumericStats::Min(rstats)) {
 			// if l.max <= r.min, it is always true ONLY if neither side contains nulls
 			return has_null ? FilterPropagateResult::FILTER_TRUE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_TRUE;
 		}
 		// if l.min > rstats.max, the filter is always false
-		if (lstats.min > rstats.max) {
+		if (NumericStats::Min(lstats) > NumericStats::Max(rstats)) {
 			return has_null ? FilterPropagateResult::FILTER_FALSE_OR_NULL : FilterPropagateResult::FILTER_ALWAYS_FALSE;
 		}
 		return FilterPropagateResult::NO_PRUNING_POSSIBLE;
@@ -103,10 +101,10 @@ unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundCompar
 	auto propagate_result = PropagateComparison(*left_stats, *right_stats, expr.type);
 	switch (propagate_result) {
 	case FilterPropagateResult::FILTER_ALWAYS_TRUE:
-		*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(true));
+		*expr_ptr = make_uniq<BoundConstantExpression>(Value::BOOLEAN(true));
 		return PropagateExpression(*expr_ptr);
 	case FilterPropagateResult::FILTER_ALWAYS_FALSE:
-		*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(false));
+		*expr_ptr = make_uniq<BoundConstantExpression>(Value::BOOLEAN(false));
 		return PropagateExpression(*expr_ptr);
 	case FilterPropagateResult::FILTER_TRUE_OR_NULL: {
 		vector<unique_ptr<Expression>> children;

package/src/duckdb/src/optimizer/statistics/expression/propagate_conjunction.cpp

@@ -2,7 +2,6 @@
 #include "duckdb/optimizer/statistics_propagator.hpp"
 #include "duckdb/planner/expression/bound_conjunction_expression.hpp"
 #include "duckdb/planner/expression/bound_constant_expression.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
 #include "duckdb/optimizer/expression_rewriter.hpp"
 #include "duckdb/execution/expression_executor.hpp"
 
@@ -51,12 +50,12 @@ unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundConjun
 			expr_idx--;
 			continue;
 		}
-		*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(constant_value));
+		*expr_ptr = make_uniq<BoundConstantExpression>(Value::BOOLEAN(constant_value));
 		return PropagateExpression(*expr_ptr);
 	}
 	if (expr.children.empty()) {
 		// if there are no children left, replace the conjunction with TRUE (for AND) or FALSE (for OR)
-		*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(is_and));
+		*expr_ptr = make_uniq<BoundConstantExpression>(Value::BOOLEAN(is_and));
 		return PropagateExpression(*expr_ptr);
 	} else if (expr.children.size() == 1) {
 		// if there is one child left, replace the conjunction with that one child

package/src/duckdb/src/optimizer/statistics/expression/propagate_constant.cpp

@@ -1,85 +1,13 @@
 #include "duckdb/optimizer/statistics_propagator.hpp"
 #include "duckdb/planner/expression/bound_constant_expression.hpp"
 #include "duckdb/storage/statistics/distinct_statistics.hpp"
-#include "duckdb/storage/statistics/list_statistics.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
-#include "duckdb/storage/statistics/string_statistics.hpp"
-#include "duckdb/storage/statistics/struct_statistics.hpp"
+#include "duckdb/storage/statistics/list_stats.hpp"
+#include "duckdb/storage/statistics/struct_stats.hpp"
 
 namespace duckdb {
 
-void UpdateDistinctStats(BaseStatistics &distinct_stats, const Value &input) {
-	Vector v(input);
-	auto &d_stats = (DistinctStatistics &)distinct_stats;
-	d_stats.Update(v, 1);
-}
-
 unique_ptr<BaseStatistics> StatisticsPropagator::StatisticsFromValue(const Value &input) {
-	switch (input.type().InternalType()) {
-	case PhysicalType::BOOL:
-	case PhysicalType::UINT8:
-	case PhysicalType::UINT16:
-	case PhysicalType::UINT32:
-	case PhysicalType::UINT64:
-	case PhysicalType::INT8:
-	case PhysicalType::INT16:
-	case PhysicalType::INT32:
-	case PhysicalType::INT64:
-	case PhysicalType::INT128:
-	case PhysicalType::FLOAT:
-	case PhysicalType::DOUBLE: {
-		auto result = make_unique<NumericStatistics>(input.type(), input, input, StatisticsType::GLOBAL_STATS);
-		result->validity_stats = make_unique<ValidityStatistics>(input.IsNull(), !input.IsNull());
-		UpdateDistinctStats(*result->distinct_stats, input);
-		return std::move(result);
-	}
-	case PhysicalType::VARCHAR: {
-		auto result = make_unique<StringStatistics>(input.type(), StatisticsType::GLOBAL_STATS);
-		result->validity_stats = make_unique<ValidityStatistics>(input.IsNull(), !input.IsNull());
-		UpdateDistinctStats(*result->distinct_stats, input);
-		if (!input.IsNull()) {
-			auto &string_value = StringValue::Get(input);
-			result->Update(string_t(string_value));
-		}
-		return std::move(result);
-	}
-	case PhysicalType::STRUCT: {
-		auto result = make_unique<StructStatistics>(input.type());
-		result->validity_stats = make_unique<ValidityStatistics>(input.IsNull(), !input.IsNull());
-		if (input.IsNull()) {
-			for (auto &child_stat : result->child_stats) {
-				child_stat.reset();
-			}
-		} else {
-			auto &struct_children = StructValue::GetChildren(input);
-			D_ASSERT(result->child_stats.size() == struct_children.size());
-			for (idx_t i = 0; i < result->child_stats.size(); i++) {
-				result->child_stats[i] = StatisticsFromValue(struct_children[i]);
-			}
-		}
-		return std::move(result);
-	}
-	case PhysicalType::LIST: {
-		auto result = make_unique<ListStatistics>(input.type());
-		result->validity_stats = make_unique<ValidityStatistics>(input.IsNull(), !input.IsNull());
-		if (input.IsNull()) {
-			result->child_stats.reset();
-		} else {
-			auto &list_children = ListValue::GetChildren(input);
-			for (auto &child_element : list_children) {
-				auto child_element_stats = StatisticsFromValue(child_element);
-				if (child_element_stats) {
-					result->child_stats->Merge(*child_element_stats);
-				} else {
-					result->child_stats.reset();
-				}
-			}
-		}
-		return std::move(result);
-	}
-	default:
-		return nullptr;
-	}
+	return BaseStatistics::FromConstant(input).ToUnique();
 }
 
 unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundConstantExpression &constant,

package/src/duckdb/src/optimizer/statistics/expression/propagate_function.cpp

@@ -5,10 +5,15 @@ namespace duckdb {
 
 unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundFunctionExpression &func,
                                                                      unique_ptr<Expression> *expr_ptr) {
-	vector<unique_ptr<BaseStatistics>> stats;
+	vector<BaseStatistics> stats;
 	stats.reserve(func.children.size());
 	for (idx_t i = 0; i < func.children.size(); i++) {
-		stats.push_back(PropagateExpression(func.children[i]));
+		auto stat = PropagateExpression(func.children[i]);
+		if (!stat) {
+			stats.push_back(BaseStatistics::CreateUnknown(func.children[i]->return_type));
+		} else {
+			stats.push_back(stat->Copy());
+		}
 	}
 	if (!func.function.statistics) {
 		return nullptr;

package/src/duckdb/src/optimizer/statistics/expression/propagate_operator.cpp

@@ -59,14 +59,24 @@ unique_ptr<BaseStatistics> StatisticsPropagator::PropagateExpression(BoundOperat
 	case ExpressionType::OPERATOR_IS_NULL:
 		if (!child_stats[0]->CanHaveNull()) {
 			// child has no null values: x IS NULL will always be false
-			*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(false));
+			*expr_ptr = make_uniq<BoundConstantExpression>(Value::BOOLEAN(false));
+			return PropagateExpression(*expr_ptr);
+		}
+		if (!child_stats[0]->CanHaveNoNull()) {
+			// child has no valid values: x IS NULL will always be true
+			*expr_ptr = make_uniq<BoundConstantExpression>(Value::BOOLEAN(true));
 			return PropagateExpression(*expr_ptr);
 		}
 		return nullptr;
 	case ExpressionType::OPERATOR_IS_NOT_NULL:
 		if (!child_stats[0]->CanHaveNull()) {
 			// child has no null values: x IS NOT NULL will always be true
-			*expr_ptr = make_unique<BoundConstantExpression>(Value::BOOLEAN(true));
+			*expr_ptr = make_uniq<BoundConstantExpression>(Value::BOOLEAN(true));
+			return PropagateExpression(*expr_ptr);
+		}
+		if (!child_stats[0]->CanHaveNoNull()) {
+			// child has no valid values: x IS NOT NULL will always be false
+			*expr_ptr = make_uniq<BoundConstantExpression>(Value::BOOLEAN(false));
 			return PropagateExpression(*expr_ptr);
 		}
 		return nullptr;

package/src/duckdb/src/optimizer/statistics/operator/propagate_aggregate.cpp

@@ -1,6 +1,5 @@
 #include "duckdb/optimizer/statistics_propagator.hpp"
 #include "duckdb/planner/operator/logical_aggregate.hpp"
-#include "duckdb/storage/statistics/validity_statistics.hpp"
 
 namespace duckdb {
 
@@ -13,14 +12,14 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalAggr
 	aggr.group_stats.resize(aggr.groups.size());
 	for (idx_t group_idx = 0; group_idx < aggr.groups.size(); group_idx++) {
 		auto stats = PropagateExpression(aggr.groups[group_idx]);
-		aggr.group_stats[group_idx] = stats ? stats->Copy() : nullptr;
+		aggr.group_stats[group_idx] = stats ? stats->ToUnique() : nullptr;
 		if (!stats) {
 			continue;
 		}
 		if (aggr.grouping_sets.size() > 1) {
 			// aggregates with multiple grouping sets can introduce NULL values to certain groups
 			// FIXME: actually figure out WHICH groups can have null values introduced
-			stats->validity_stats = make_unique<ValidityStatistics>(true, true);
+			stats->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 			continue;
 		}
 		ColumnBinding group_binding(aggr.group_index, group_idx);

package/src/duckdb/src/optimizer/statistics/operator/propagate_filter.cpp

@@ -4,8 +4,9 @@
 #include "duckdb/planner/expression/bound_columnref_expression.hpp"
 #include "duckdb/planner/expression/bound_comparison_expression.hpp"
 #include "duckdb/planner/expression/bound_constant_expression.hpp"
+#include "duckdb/planner/expression/bound_function_expression.hpp"
 #include "duckdb/planner/operator/logical_filter.hpp"
-#include "duckdb/storage/statistics/numeric_statistics.hpp"
+#include "duckdb/storage/statistics/base_statistics.hpp"
 
 namespace duckdb {
 
@@ -17,7 +18,7 @@ bool StatisticsPropagator::ExpressionIsConstant(Expression &expr, const Value &v
 	if (expr.GetExpressionClass() != ExpressionClass::BOUND_CONSTANT) {
 		return false;
 	}
-	auto &bound_constant = (BoundConstantExpression &)expr;
+	auto &bound_constant = expr.Cast<BoundConstantExpression>();
 	D_ASSERT(bound_constant.value.type() == val.type());
 	return Value::NotDistinctFrom(bound_constant.value, val);
 }
@@ -26,7 +27,7 @@ bool StatisticsPropagator::ExpressionIsConstantOrNull(Expression &expr, const Va
 	if (expr.GetExpressionClass() != ExpressionClass::BOUND_FUNCTION) {
 		return false;
 	}
-	auto &bound_function = (BoundFunctionExpression &)expr;
+	auto &bound_function = expr.Cast<BoundFunctionExpression>();
 	return ConstantOrNull::IsConstantOrNull(bound_function, val);
 }
 
@@ -35,21 +36,20 @@ void StatisticsPropagator::SetStatisticsNotNull(ColumnBinding binding) {
 	if (entry == statistics_map.end()) {
 		return;
 	}
-	entry->second->validity_stats = make_unique<ValidityStatistics>(false);
+	entry->second->Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
 }
 
 void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &stats, ExpressionType comparison_type,
                                                   const Value &constant) {
 	// regular comparisons removes all null values
 	if (!IsCompareDistinct(comparison_type)) {
-		stats.validity_stats = make_unique<ValidityStatistics>(false);
+		stats.Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
 	}
-	if (!stats.type.IsNumeric()) {
+	if (!stats.GetType().IsNumeric()) {
 		// don't handle non-numeric columns here (yet)
 		return;
 	}
-	auto &numeric_stats = (NumericStatistics &)stats;
-	if (numeric_stats.min.IsNull() || numeric_stats.max.IsNull()) {
+	if (!NumericStats::HasMinMax(stats)) {
 		// no stats available: skip this
 		return;
 	}
@@ -58,19 +58,19 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &stats, Express
 	case ExpressionType::COMPARE_LESSTHANOREQUALTO:
 		// X < constant OR X <= constant
 		// max becomes the constant
-		numeric_stats.max = constant;
+		NumericStats::SetMax(stats, constant);
 		break;
 	case ExpressionType::COMPARE_GREATERTHAN:
 	case ExpressionType::COMPARE_GREATERTHANOREQUALTO:
 		// X > constant OR X >= constant
 		// min becomes the constant
-		numeric_stats.min = constant;
+		NumericStats::SetMin(stats, constant);
 		break;
 	case ExpressionType::COMPARE_EQUAL:
 		// X = constant
 		// both min and max become the constant
-		numeric_stats.min = constant;
-		numeric_stats.max = constant;
+		NumericStats::SetMin(stats, constant);
+		NumericStats::SetMax(stats, constant);
 		break;
 	default:
 		break;
@@ -81,17 +81,15 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &lstats, BaseSt
                                                   ExpressionType comparison_type) {
 	// regular comparisons removes all null values
 	if (!IsCompareDistinct(comparison_type)) {
-		lstats.validity_stats = make_unique<ValidityStatistics>(false);
-		rstats.validity_stats = make_unique<ValidityStatistics>(false);
+		lstats.Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
+		rstats.Set(StatsInfo::CANNOT_HAVE_NULL_VALUES);
 	}
-	D_ASSERT(lstats.type == rstats.type);
-	if (!lstats.type.IsNumeric()) {
+	D_ASSERT(lstats.GetType() == rstats.GetType());
+	if (!lstats.GetType().IsNumeric()) {
 		// don't handle non-numeric columns here (yet)
 		return;
 	}
-	auto &left_stats = (NumericStatistics &)lstats;
-	auto &right_stats = (NumericStatistics &)rstats;
-	if (left_stats.min.IsNull() || left_stats.max.IsNull() || right_stats.min.IsNull() || right_stats.max.IsNull()) {
+	if (!NumericStats::HasMinMax(lstats) || !NumericStats::HasMinMax(rstats)) {
 		// no stats available: skip this
 		return;
 	}
@@ -104,14 +102,14 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &lstats, BaseSt
 
 		// we know that left.max is AT MOST equal to right.max
 		// because any value in left that is BIGGER than right.max will not pass the filter
-		if (left_stats.max > right_stats.max) {
-			left_stats.max = right_stats.max;
+		if (NumericStats::Max(lstats) > NumericStats::Max(rstats)) {
+			NumericStats::SetMax(lstats, NumericStats::Max(rstats));
 		}
 
 		// we also know that right.min is AT MOST equal to left.min
 		// because any value in right that is SMALLER than left.min will not pass the filter
-		if (right_stats.min < left_stats.min) {
-			right_stats.min = left_stats.min;
+		if (NumericStats::Min(rstats) < NumericStats::Min(lstats)) {
+			NumericStats::SetMin(rstats, NumericStats::Min(lstats));
 		}
 		// so in our example, the bounds get updated as follows:
 		// left: [-50, 100], right: [-50, 100]
@@ -121,11 +119,11 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &lstats, BaseSt
 		// LEFT > RIGHT OR LEFT >= RIGHT
 		// we know that every value of left is bigger (or equal to) every value in right
 		// this is essentially the inverse of the less than (or equal to) scenario
-		if (right_stats.max > left_stats.max) {
-			right_stats.max = left_stats.max;
+		if (NumericStats::Max(rstats) > NumericStats::Max(lstats)) {
+			NumericStats::SetMax(rstats, NumericStats::Max(lstats));
 		}
-		if (left_stats.min < right_stats.min) {
-			left_stats.min = right_stats.min;
+		if (NumericStats::Min(lstats) < NumericStats::Min(rstats)) {
+			NumericStats::SetMin(lstats, NumericStats::Min(rstats));
 		}
 		break;
 	case ExpressionType::COMPARE_EQUAL:
@@ -135,16 +133,16 @@ void StatisticsPropagator::UpdateFilterStatistics(BaseStatistics &lstats, BaseSt
 		// so if we have e.g. left = [-50, 250] and right = [-100, 100]
 		// the tighest bounds are [-50, 100]
 		// select the highest min
-		if (left_stats.min > right_stats.min) {
-			right_stats.min = left_stats.min;
+		if (NumericStats::Min(lstats) > NumericStats::Min(rstats)) {
+			NumericStats::SetMin(rstats, NumericStats::Min(lstats));
 		} else {
-			left_stats.min = right_stats.min;
+			NumericStats::SetMin(lstats, NumericStats::Min(rstats));
 		}
 		// select the lowest max
-		if (left_stats.max < right_stats.max) {
-			right_stats.max = left_stats.max;
+		if (NumericStats::Max(lstats) < NumericStats::Max(rstats)) {
+			NumericStats::SetMax(rstats, NumericStats::Max(lstats));
 		} else {
-			left_stats.max = right_stats.max;
+			NumericStats::SetMax(lstats, NumericStats::Max(rstats));
 		}
 		break;
 	default:
@@ -157,10 +155,10 @@ void StatisticsPropagator::UpdateFilterStatistics(Expression &left, Expression &
 	// any column ref involved in a comparison will not be null after the comparison
 	bool compare_distinct = IsCompareDistinct(comparison_type);
 	if (!compare_distinct && left.type == ExpressionType::BOUND_COLUMN_REF) {
-		SetStatisticsNotNull(((BoundColumnRefExpression &)left).binding);
+		SetStatisticsNotNull((left.Cast<BoundColumnRefExpression>()).binding);
 	}
 	if (!compare_distinct && right.type == ExpressionType::BOUND_COLUMN_REF) {
-		SetStatisticsNotNull(((BoundColumnRefExpression &)right).binding);
+		SetStatisticsNotNull((right.Cast<BoundColumnRefExpression>()).binding);
 	}
 	// check if this is a comparison between a constant and a column ref
 	BoundConstantExpression *constant = nullptr;
@@ -168,14 +166,14 @@ void StatisticsPropagator::UpdateFilterStatistics(Expression &left, Expression &
 	if (left.type == ExpressionType::VALUE_CONSTANT && right.type == ExpressionType::BOUND_COLUMN_REF) {
 		constant = (BoundConstantExpression *)&left;
 		columnref = (BoundColumnRefExpression *)&right;
-		comparison_type = FlipComparisionExpression(comparison_type);
+		comparison_type = FlipComparisonExpression(comparison_type);
 	} else if (left.type == ExpressionType::BOUND_COLUMN_REF && right.type == ExpressionType::VALUE_CONSTANT) {
 		columnref = (BoundColumnRefExpression *)&left;
 		constant = (BoundConstantExpression *)&right;
 	} else if (left.type == ExpressionType::BOUND_COLUMN_REF && right.type == ExpressionType::BOUND_COLUMN_REF) {
 		// comparison between two column refs
-		auto &left_column_ref = (BoundColumnRefExpression &)left;
-		auto &right_column_ref = (BoundColumnRefExpression &)right;
+		auto &left_column_ref = left.Cast<BoundColumnRefExpression>();
+		auto &right_column_ref = right.Cast<BoundColumnRefExpression>();
 		auto lentry = statistics_map.find(left_column_ref.binding);
 		auto rentry = statistics_map.find(right_column_ref.binding);
 		if (lentry == statistics_map.end() || rentry == statistics_map.end()) {
@@ -201,13 +199,13 @@ void StatisticsPropagator::UpdateFilterStatistics(Expression &condition) {
 	// if we find a comparison in the form of e.g. "i=3", we can update our statistics for that column
 	switch (condition.GetExpressionClass()) {
 	case ExpressionClass::BOUND_BETWEEN: {
-		auto &between = (BoundBetweenExpression &)condition;
+		auto &between = condition.Cast<BoundBetweenExpression>();
 		UpdateFilterStatistics(*between.input, *between.lower, between.LowerComparisonType());
 		UpdateFilterStatistics(*between.input, *between.upper, between.UpperComparisonType());
 		break;
 	}
 	case ExpressionClass::BOUND_COMPARISON: {
-		auto &comparison = (BoundComparisonExpression &)condition;
+		auto &comparison = condition.Cast<BoundComparisonExpression>();
 		UpdateFilterStatistics(*comparison.left, *comparison.right, comparison.type);
 		break;
 	}
@@ -222,7 +220,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalFilt
 	node_stats = PropagateStatistics(filter.children[0]);
 	if (filter.children[0]->type == LogicalOperatorType::LOGICAL_EMPTY_RESULT) {
 		ReplaceWithEmptyResult(*node_ptr);
-		return make_unique<NodeStatistics>(0, 0);
+		return make_uniq<NodeStatistics>(0, 0);
 	}
 
 	// then propagate to each of the expressions
@@ -244,7 +242,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalFilt
 		           ExpressionIsConstantOrNull(*condition, Value::BOOLEAN(false))) {
 			// filter is always false or null; this entire filter should be replaced by an empty result block
 			ReplaceWithEmptyResult(*node_ptr);
-			return make_unique<NodeStatistics>(0, 0);
+			return make_uniq<NodeStatistics>(0, 0);
 		} else {
 			// cannot prune this filter: propagate statistics from the filter
 			UpdateFilterStatistics(*condition);

package/src/duckdb/src/optimizer/statistics/operator/propagate_get.cpp

@@ -86,7 +86,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalGet
 		case FilterPropagateResult::FILTER_ALWAYS_FALSE:
 			// filter is always false; this entire filter should be replaced by an empty result block
 			ReplaceWithEmptyResult(*node_ptr);
-			return make_unique<NodeStatistics>(0, 0);
+			return make_uniq<NodeStatistics>(0, 0);
 		default:
 			// general case: filter can be true or false, update this columns' statistics
 			UpdateFilterStatistics(stats, *filter);

package/src/duckdb/src/optimizer/statistics/operator/propagate_join.cpp

@@ -7,7 +7,6 @@
 #include "duckdb/planner/operator/logical_join.hpp"
 #include "duckdb/planner/operator/logical_limit.hpp"
 #include "duckdb/planner/operator/logical_positional_join.hpp"
-#include "duckdb/storage/statistics/validity_statistics.hpp"
 
 namespace duckdb {
 
@@ -41,7 +40,7 @@ void StatisticsPropagator::PropagateStatistics(LogicalComparisonJoin &join, uniq
 				case JoinType::ANTI: {
 					// when the right child has data, return the left child
 					// when the right child has no data, return an empty set
-					auto limit = make_unique<LogicalLimit>(1, 0, nullptr, nullptr);
+					auto limit = make_uniq<LogicalLimit>(1, 0, nullptr, nullptr);
 					limit->AddChild(std::move(join.children[1]));
 					auto cross_product = LogicalCrossProduct::Create(std::move(join.children[0]), std::move(limit));
 					*node_ptr = std::move(cross_product);
@@ -77,7 +76,7 @@ void StatisticsPropagator::PropagateStatistics(LogicalComparisonJoin &join, uniq
 					case JoinType::SEMI: {
 						// when the right child has data, return the left child
 						// when the right child has no data, return an empty set
-						auto limit = make_unique<LogicalLimit>(1, 0, nullptr, nullptr);
+						auto limit = make_uniq<LogicalLimit>(1, 0, nullptr, nullptr);
 						limit->AddChild(std::move(join.children[1]));
 						auto cross_product = LogicalCrossProduct::Create(std::move(join.children[0]), std::move(limit));
 						*node_ptr = std::move(cross_product);
@@ -196,10 +195,11 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalJoin
 	switch (join.type) {
 	case LogicalOperatorType::LOGICAL_COMPARISON_JOIN:
 	case LogicalOperatorType::LOGICAL_DELIM_JOIN:
-		PropagateStatistics((LogicalComparisonJoin &)join, node_ptr);
+	case LogicalOperatorType::LOGICAL_ASOF_JOIN:
+		PropagateStatistics(join.Cast<LogicalComparisonJoin>(), node_ptr);
 		break;
 	case LogicalOperatorType::LOGICAL_ANY_JOIN:
-		PropagateStatistics((LogicalAnyJoin &)join, node_ptr);
+		PropagateStatistics(join.Cast<LogicalAnyJoin>(), node_ptr);
 		break;
 	default:
 		break;
@@ -210,7 +210,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalJoin
 		for (auto &binding : right_bindings) {
 			auto stats = statistics_map.find(binding);
 			if (stats != statistics_map.end()) {
-				stats->second->validity_stats = make_unique<ValidityStatistics>(true);
+				stats->second->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 			}
 		}
 	}
@@ -219,7 +219,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalJoin
 		for (auto &binding : left_bindings) {
 			auto stats = statistics_map.find(binding);
 			if (stats != statistics_map.end()) {
-				stats->second->validity_stats = make_unique<ValidityStatistics>(true);
+				stats->second->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 			}
 		}
 	}
@@ -265,7 +265,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalPosi
 	for (auto &binding : left_bindings) {
 		auto stats = statistics_map.find(binding);
 		if (stats != statistics_map.end()) {
-			stats->second->validity_stats = make_unique<ValidityStatistics>(true);
+			stats->second->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 		}
 	}
 
@@ -274,7 +274,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalPosi
 	for (auto &binding : right_bindings) {
 		auto stats = statistics_map.find(binding);
 		if (stats != statistics_map.end()) {
-			stats->second->validity_stats = make_unique<ValidityStatistics>(true);
+			stats->second->Set(StatsInfo::CAN_HAVE_NULL_VALUES);
 		}
 	}
 

package/src/duckdb/src/optimizer/statistics/operator/propagate_limit.cpp

@@ -8,7 +8,7 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalLimi
 	// propagate statistics in the child node
 	PropagateStatistics(limit.children[0]);
 	// return the node stats, with as expected cardinality the amount specified in the limit
-	return make_unique<NodeStatistics>(limit.limit_val, limit.limit_val);
+	return make_uniq<NodeStatistics>(limit.limit_val, limit.limit_val);
 }
 
 } // namespace duckdb

package/src/duckdb/src/optimizer/statistics/operator/propagate_set_operation.cpp

@@ -47,18 +47,18 @@ unique_ptr<NodeStatistics> StatisticsPropagator::PropagateStatistics(LogicalSetO
 		switch (setop.type) {
 		case LogicalOperatorType::LOGICAL_UNION:
 			// union: merge the stats of the LHS and RHS together
-			new_stats = left_entry->second->Copy();
+			new_stats = left_entry->second->ToUnique();
 			new_stats->Merge(*right_entry->second);
 			break;
 		case LogicalOperatorType::LOGICAL_EXCEPT:
 			// except: use the stats of the LHS
-			new_stats = left_entry->second->Copy();
+			new_stats = left_entry->second->ToUnique();
 			break;
 		case LogicalOperatorType::LOGICAL_INTERSECT:
 			// intersect: intersect the two stats
 			// FIXME: for now we just use the stats of the LHS, as this is correct
 			// however, the stats can be further refined to the minimal subset of the LHS and RHS
-			new_stats = left_entry->second->Copy();
+			new_stats = left_entry->second->ToUnique();
 			break;
 		default:
 			throw InternalException("Unsupported setop type");