Removed optional cost from statistics

Removed optional cost from statistics

10 files changed, 20 insertions, 62 deletions

diff --git a/ydb/core/kqp/opt/kqp_query_plan.cpp b/ydb/core/kqp/opt/kqp_query_plan.cpp
index 2aa2a17de8..cd4498eb02 100644
--- a/ydb/core/kqp/opt/kqp_query_plan.cpp
+++ b/ydb/core/kqp/opt/kqp_query_plan.cpp
@@ -1265,7 +1265,7 @@ private:
         
         if (auto stats = SerializerCtx.TypeCtx.GetStats(expr.Raw())) {
             op.Properties["E-Rows"] = stats->Nrows;
-            op.Properties["E-Cost"] = stats->Cost.value();
+            op.Properties["E-Cost"] = stats->Cost;
         }
         else {
             op.Properties["E-Rows"] = "No estimate";
diff --git a/ydb/core/kqp/opt/kqp_statistics_transformer.cpp b/ydb/core/kqp/opt/kqp_statistics_transformer.cpp
index 5680bc3c83..dc95487448 100644
--- a/ydb/core/kqp/opt/kqp_statistics_transformer.cpp
+++ b/ydb/core/kqp/opt/kqp_statistics_transformer.cpp
@@ -118,7 +118,7 @@ void InferStatisticsForRowsSourceSettings(const TExprNode::TPtr& input, TTypeAnn
     }
 
     int nAttrs = sourceSettings.Columns().Size();
-    double cost = inputStats->Cost.value();
+    double cost = inputStats->Cost;
 
     auto outputStats = TOptimizerStatistics(nRows, nAttrs, cost);
     typeCtx->SetStats(input.Get(), std::make_shared<TOptimizerStatistics>(outputStats));
diff --git a/ydb/library/yql/core/peephole_opt/yql_opt_peephole_physical.cpp b/ydb/library/yql/core/peephole_opt/yql_opt_peephole_physical.cpp
index 41edddac10..0027feaef4 100644
--- a/ydb/library/yql/core/peephole_opt/yql_opt_peephole_physical.cpp
+++ b/ydb/library/yql/core/peephole_opt/yql_opt_peephole_physical.cpp
@@ -5836,7 +5836,7 @@ TExprNode::TPtr ExpandCostsOf(const TExprNode::TPtr& node, TExprContext& ctx, TT
             }
             if (stat) {
                 if (stat->Cost) {
-                    jsonWriter.Write("Cost", *stat->Cost);
+                    jsonWriter.Write("Cost", stat->Cost);
                 }
                 jsonWriter.Write("Cols", stat->Ncols);
                 jsonWriter.Write("Rows", stat->Nrows);
diff --git a/ydb/library/yql/core/yql_cost_function.cpp b/ydb/library/yql/core/yql_cost_function.cpp
index 635b0552eb..f42c95164b 100644
--- a/ydb/library/yql/core/yql_cost_function.cpp
+++ b/ydb/library/yql/core/yql_cost_function.cpp
@@ -16,7 +16,7 @@ TOptimizerStatistics NYql::ComputeJoinStats(TOptimizerStatistics leftStats, TOpt
     int newNCols = leftStats.Ncols + rightStats.Ncols;
     double cost = leftStats.Nrows + 2.0 * rightStats.Nrows 
             + newCard 
-            + leftStats.Cost.value() + rightStats.Cost.value();
+            + leftStats.Cost + rightStats.Cost;
 
     return TOptimizerStatistics(newCard, newNCols, cost);
 }
diff --git a/ydb/library/yql/core/yql_statistics.cpp b/ydb/library/yql/core/yql_statistics.cpp
index 91043b31b7..30d8dcd26b 100644
--- a/ydb/library/yql/core/yql_statistics.cpp
+++ b/ydb/library/yql/core/yql_statistics.cpp
@@ -3,13 +3,7 @@
 using namespace NYql;
 
 std::ostream& NYql::operator<<(std::ostream& os, const TOptimizerStatistics& s) {
-    os << "Nrows: " << s.Nrows << ", Ncols: " << s.Ncols;
-    os << ", Cost: ";
-    if (s.Cost.has_value()){
-        os << s.Cost.value();
-    } else {
-        os << "none";
-    }
+    os << "Nrows: " << s.Nrows << ", Ncols: " << s.Ncols << ", Cost: " << s.Cost ;
     return os;
 }
 
@@ -20,10 +14,6 @@ bool TOptimizerStatistics::Empty() const {
 TOptimizerStatistics& TOptimizerStatistics::operator+=(const TOptimizerStatistics& other) {
     Nrows += other.Nrows;
     Ncols += other.Ncols;
-    if (Cost.has_value() && other.Cost.has_value()) {
-        Cost = *Cost + *other.Cost;
-    } else if (other.Cost.has_value()) {
-        Cost = other.Cost;
-    }
+    Cost += other.Cost;
     return *this;
 }
diff --git a/ydb/library/yql/core/yql_statistics.h b/ydb/library/yql/core/yql_statistics.h
index a8fed920d6..7564a751a1 100644
--- a/ydb/library/yql/core/yql_statistics.h
+++ b/ydb/library/yql/core/yql_statistics.h
@@ -11,16 +11,16 @@ namespace NYql {
  * for the current operator in the plan. Currently, only Nrows and Ncols are
  * recorded.
  * Cost is also included in statistics, as its updated concurrently with statistics
- * all of the time. Cost is optional, so it could be missing.
+ * all of the time.
 */
 struct TOptimizerStatistics {
     double Nrows = 0;
     int Ncols = 0;
-    std::optional<double> Cost;
+    double Cost;
     TString Descr;
 
-    TOptimizerStatistics() : Cost(std::nullopt) {}
-    TOptimizerStatistics(double nrows,int ncols): Nrows(nrows), Ncols(ncols), Cost(std::nullopt) {}
+    TOptimizerStatistics() {}
+    TOptimizerStatistics(double nrows,int ncols): Nrows(nrows), Ncols(ncols) {}
     TOptimizerStatistics(double nrows,int ncols, double cost): Nrows(nrows), Ncols(ncols), Cost(cost) {}
     TOptimizerStatistics(double nrows,int ncols, double cost, TString descr): Nrows(nrows), Ncols(ncols), Cost(cost), Descr(descr) {}
 
diff --git a/ydb/library/yql/core/yql_type_annotation.h b/ydb/library/yql/core/yql_type_annotation.h
index b8efdc52a6..de931c1034 100644
--- a/ydb/library/yql/core/yql_type_annotation.h
+++ b/ydb/library/yql/core/yql_type_annotation.h
@@ -338,21 +338,7 @@ struct TTypeAnnotationContext: public TThrRefBase {
     void SetStats(const TExprNode* input, std::shared_ptr<TOptimizerStatistics> stats) {
         StatisticsMap[input] = stats;
     }
-
-    /**
-     * Helper method to get cost from type annotation context
-     * Doesn't check if the cost is in the mapping
-     */
-    std::optional<double> GetCost(const TExprNode* input) {
-        return StatisticsMap[input]->Cost;
-    }
-
-    /**
-     * Helper method to set the cost in type annotation context
-     */
-    void SetCost(const TExprNode* input, std::optional<double> cost) {
-        StatisticsMap[input]->Cost = cost;
-    }
+    
 };
 
 template <> inline
diff --git a/ydb/library/yql/dq/opt/dq_opt_join_cost_based.cpp b/ydb/library/yql/dq/opt/dq_opt_join_cost_based.cpp
index b9d74d1877..343d7704fb 100644
--- a/ydb/library/yql/dq/opt/dq_opt_join_cost_based.cpp
+++ b/ydb/library/yql/dq/opt/dq_opt_join_cost_based.cpp
@@ -665,7 +665,7 @@ template <int N> void TDPccpSolver<N>::EmitCsgCmp(const std::bitset<N>& S1, cons
         TEdge e2 = Graph.FindCrossingEdge(S2, S1);
         std::shared_ptr<TJoinOptimizerNode> newJoin = 
             MakeJoin(DpTable[S2], DpTable[S1], e2.JoinConditions, GraceJoin);
-        if (newJoin->Stats->Cost.value() < DpTable[joined]->Stats->Cost.value()){
+        if (newJoin->Stats->Cost < DpTable[joined]->Stats->Cost){
             DpTable[joined] = newJoin;
         }
     } else {
@@ -675,10 +675,10 @@ template <int N> void TDPccpSolver<N>::EmitCsgCmp(const std::bitset<N>& S1, cons
         TEdge e2 = Graph.FindCrossingEdge(S2, S1);
         std::shared_ptr<TJoinOptimizerNode> newJoin2 = 
             MakeJoin(DpTable[S2], DpTable[S1], e2.JoinConditions, GraceJoin);
-        if (newJoin1->Stats->Cost.value() < DpTable[joined]->Stats->Cost.value()){
+        if (newJoin1->Stats->Cost < DpTable[joined]->Stats->Cost){
             DpTable[joined] = newJoin1;
         }
-        if (newJoin2->Stats->Cost.value() < DpTable[joined]->Stats->Cost.value()){
+        if (newJoin2->Stats->Cost < DpTable[joined]->Stats->Cost){
             DpTable[joined] = newJoin2;
         }
     }
@@ -860,10 +860,6 @@ bool DqCollectJoinRelationsWithStats(
             return false;
         }
 
-        if (!maybeStat->second->Cost.has_value()) {
-            return false;
-        }
-
         auto scope = input.Scope();
         if (!scope.Maybe<TCoAtom>()){
             return false;
@@ -899,8 +895,7 @@ TExprBase DqOptimizeEquiJoinWithCosts(const TExprBase& node, TExprContext& ctx,
     auto equiJoin = node.Cast<TCoEquiJoin>();
     YQL_ENSURE(equiJoin.ArgCount() >= 4);
 
-    if (typesCtx.StatisticsMap.contains(equiJoin.Raw()) && 
-        typesCtx.StatisticsMap[equiJoin.Raw()]->Cost.has_value()) {
+    if (typesCtx.StatisticsMap.contains(equiJoin.Raw())) {
 
         return node;
     }
@@ -999,7 +994,7 @@ public:
         TVector<int> scope;
         BuildOutput(&output, result.get(), scope);
         output.Rows = result->Stats->Nrows;
-        output.TotalCost = *result->Stats->Cost;
+        output.TotalCost = result->Stats->Cost;
         if (Log) {
             Log(output.ToString());
         }
diff --git a/ydb/library/yql/dq/opt/dq_opt_join_cost_based_generic.cpp b/ydb/library/yql/dq/opt/dq_opt_join_cost_based_generic.cpp
index 27c05b4e1c..2f9c42d02a 100644
--- a/ydb/library/yql/dq/opt/dq_opt_join_cost_based_generic.cpp
+++ b/ydb/library/yql/dq/opt/dq_opt_join_cost_based_generic.cpp
@@ -44,7 +44,7 @@ struct TState {
         VarIds.emplace_back();
         int relId = Input.Rels.size();
         Input.Rels.back().Rows = stat->Nrows;
-        Input.Rels.back().TotalCost = *stat->Cost;
+        Input.Rels.back().TotalCost = stat->Cost;
         Tables.emplace_back(label);
         Table2RelIds[label].emplace_back(relId);
     }
@@ -171,8 +171,7 @@ TExprBase DqOptimizeEquiJoinWithCosts(
     YQL_ENSURE(equiJoin.ArgCount() >= 4);
 
 	auto maybeStat = typesCtx.StatisticsMap.find(equiJoin.Raw());
-    if (maybeStat != typesCtx.StatisticsMap.end() &&
-        maybeStat->second->Cost.has_value()) {
+    if (maybeStat != typesCtx.StatisticsMap.end()) {
         return node;
     }
 
diff --git a/ydb/library/yql/dq/opt/dq_opt_stat.cpp b/ydb/library/yql/dq/opt/dq_opt_stat.cpp
index 9e14bf142e..3e838da4b9 100644
--- a/ydb/library/yql/dq/opt/dq_opt_stat.cpp
+++ b/ydb/library/yql/dq/opt/dq_opt_stat.cpp
@@ -67,7 +67,6 @@ void InferStatisticsForDqSource(const TExprNode::TPtr& input, TTypeAnnotationCon
     }
 
     typeCtx->SetStats(input.Get(), inputStats);
-    typeCtx->SetCost(input.Get(), typeCtx->GetCost(dqSource.Settings().Raw()));
 }
 
 /**
@@ -95,11 +94,9 @@ void InferStatisticsForFlatMap(const TExprNode::TPtr& input, TTypeAnnotationCont
 
         double selectivity = ComputePredicateSelectivity(flatmap.Lambda().Body(), inputStats);
 
-        auto outputStats = TOptimizerStatistics(inputStats->Nrows * selectivity, inputStats->Ncols);
+        auto outputStats = TOptimizerStatistics(inputStats->Nrows * selectivity, inputStats->Ncols, inputStats->Cost );
 
         typeCtx->SetStats(input.Get(), std::make_shared<TOptimizerStatistics>(outputStats) );
-        typeCtx->SetCost(input.Get(), typeCtx->GetCost(flatmapInput.Raw()));
-
     }
     else if (flatmap.Lambda().Body().Maybe<TCoMapJoinCore>() || 
             flatmap.Lambda().Body().Maybe<TCoMap>().Input().Maybe<TCoMapJoinCore>() ||
@@ -138,11 +135,9 @@ void InferStatisticsForFilter(const TExprNode::TPtr& input, TTypeAnnotationConte
     
     double selectivity = ComputePredicateSelectivity(filterBody, inputStats);
 
-    auto outputStats = TOptimizerStatistics(inputStats->Nrows * selectivity, inputStats->Ncols);
+    auto outputStats = TOptimizerStatistics(inputStats->Nrows * selectivity, inputStats->Ncols, inputStats->Cost);
 
     typeCtx->SetStats(input.Get(), std::make_shared<TOptimizerStatistics>(outputStats) );
-    typeCtx->SetCost(input.Get(), typeCtx->GetCost(filterInput.Raw()));
-
 }
 
 /**
@@ -162,7 +157,6 @@ void InferStatisticsForSkipNullMembers(const TExprNode::TPtr& input, TTypeAnnota
     }
 
     typeCtx->SetStats( input.Get(), inputStats );
-    typeCtx->SetCost( input.Get(), typeCtx->GetCost( skipNullMembersInput.Raw() ) );
 }
 
 /**
@@ -181,7 +175,6 @@ void InferStatisticsForExtractMembers(const TExprNode::TPtr& input, TTypeAnnotat
     }
 
     typeCtx->SetStats( input.Get(), inputStats );
-    typeCtx->SetCost( input.Get(), typeCtx->GetCost( extractMembersInput.Raw() ) );
 }
 
 /**
@@ -200,7 +193,6 @@ void InferStatisticsForAggregateCombine(const TExprNode::TPtr& input, TTypeAnnot
     }
 
     typeCtx->SetStats( input.Get(), inputStats );
-    typeCtx->SetCost( input.Get(), typeCtx->GetCost( aggInput.Raw() ) );
 }
 
 /**
@@ -219,7 +211,6 @@ void InferStatisticsForAggregateMergeFinalize(const TExprNode::TPtr& input, TTyp
     }
 
     typeCtx->SetStats( input.Get(), inputStats );
-    typeCtx->SetCost( input.Get(), typeCtx->GetCost( aggInput.Raw() ) );
 }
 
 /***
@@ -255,7 +246,6 @@ void PropagateStatisticsToLambdaArgument(const TExprNode::TPtr& input, TTypeAnno
                 auto inputStats = typeCtx->GetStats(callableInput->Child(j) );
                 if (inputStats){
                     typeCtx->SetStats( lambda.Args().Arg(j).Raw(), inputStats );
-                    typeCtx->SetCost( lambda.Args().Arg(j).Raw(), typeCtx->GetCost( callableInput->Child(j) ));
                 }
             }
             
@@ -267,7 +257,6 @@ void PropagateStatisticsToLambdaArgument(const TExprNode::TPtr& input, TTypeAnno
             }
 
             typeCtx->SetStats( lambda.Args().Arg(0).Raw(), inputStats );
-            typeCtx->SetCost( lambda.Args().Arg(0).Raw(), typeCtx->GetCost( callableInput.Get() ));
         }
     }
 }
@@ -282,7 +271,6 @@ void InferStatisticsForStage(const TExprNode::TPtr& input, TTypeAnnotationContex
     auto lambdaStats = typeCtx->GetStats( stage.Program().Body().Raw());
     if (lambdaStats){
         typeCtx->SetStats( stage.Raw(), lambdaStats );
-        typeCtx->SetCost( stage.Raw(), typeCtx->GetCost( stage.Program().Body().Raw()));
     }
 }