Fix input descs ordering to get determined UNION ALL results.

Signed-off-by: ienkovich <[email protected]>

Author: ienkovich

omniscidb/QueryEngine/RelAlgExecutor.cpp

@@ -1494,6 +1494,14 @@ ExecutionResult RelAlgExecutor::executeStep(const hdk::ir::Node* step_root,
   auto sort = step_root->as<hdk::ir::Sort>();
   ExecutionOptions eo_with_limit =
       eo.with_just_validate(eo.just_validate || (sort && sort->isEmptyResult()));
+  // Use additional result fragments sort for UNION ALL case.
+  // Detect it via a check for two outer tables in the input
+  // descriptors vector.
+  if (work_unit.exe_unit.input_descs.size() >= 2 &&
+      work_unit.exe_unit.input_descs[0].getNestLevel() ==
+          work_unit.exe_unit.input_descs[1].getNestLevel()) {
+    eo_with_limit = eo_with_limit.with_preserve_order(true);
+  }
 
   bool cpu_only = false;
   if (auto project = step_root->as<hdk::ir::Project>()) {

omniscidb/QueryEngine/WorkUnitBuilder.cpp

@@ -494,7 +494,13 @@ int WorkUnitBuilder::assignNestLevels(const ir::Node* node, int start_idx) {
 
   if (node->is<ir::LogicalUnion>()) {
     for (size_t i = 0; i < node->inputCount(); ++i) {
-      assignNestLevels(node->getInput(i), start_idx);
+      auto input_node = node->getInput(i);
+      assignNestLevels(input_node, start_idx);
+      if (auto scan = input_node->as<ir::Scan>()) {
+        union_order_[{scan->getDatabaseId(), scan->getTableId()}] = i;
+      } else {
+        union_order_[{-1, -(int)input_node->getId()}] = i;
+      }
     }
     ++start_idx;
   } else {
@@ -547,8 +553,13 @@ void WorkUnitBuilder::computeInputDescs() {
 
   std::sort(input_descs_.begin(),
             input_descs_.end(),
-            [](const InputDescriptor& lhs, const InputDescriptor& rhs) {
-              return lhs.getNestLevel() < rhs.getNestLevel();
+            [this](const InputDescriptor& lhs, const InputDescriptor& rhs) {
+              if (lhs.getNestLevel() != rhs.getNestLevel()) {
+                return lhs.getNestLevel() < rhs.getNestLevel();
+              }
+              TableRef lhs_ref{lhs.getDatabaseId(), lhs.getTableId()};
+              TableRef rhs_ref{rhs.getDatabaseId(), rhs.getTableId()};
+              return union_order_.at(lhs_ref) < union_order_.at(rhs_ref);
             });
 }
 

omniscidb/QueryEngine/WorkUnitBuilder.h

@@ -120,6 +120,9 @@ class WorkUnitBuilder {
   std::unordered_map<const ir::Node*, int> input_nest_levels_;
   // Stores nest level for each node to process.
   std::unordered_map<const ir::Node*, int> all_nest_levels_;
+  // Stores order for each UNION ALL input node for proper
+  // result ordering.
+  std::unordered_map<TableRef, int> union_order_;
   std::vector<JoinType> join_types_;
   bool is_agg_ = false;