Do nut use Arena in ColumnsHashing for string in functions arrayUniq and arrayEnumerate.

dbms/src/Common/ColumnsHashing.h

@@ -62,11 +62,11 @@ struct HashMethodOneNumber
 
 
 /// For the case where there is one string key.
-template <typename Value, typename Mapped, bool use_cache = true>
+template <typename Value, typename Mapped, bool place_string_to_arena = true, bool use_cache = true>
 struct HashMethodString
-    : public columns_hashing_impl::HashMethodBase<HashMethodString<Value, Mapped, use_cache>, Value, Mapped, use_cache>
+    : public columns_hashing_impl::HashMethodBase<HashMethodString<Value, Mapped, place_string_to_arena, use_cache>, Value, Mapped, use_cache>
 {
-    using Self = HashMethodString<Value, Mapped, use_cache>;
+    using Self = HashMethodString<Value, Mapped, place_string_to_arena, use_cache>;
     using Base = columns_hashing_impl::HashMethodBase<Self, Value, Mapped, use_cache>;
 
     const IColumn::Offset * offsets;
@@ -90,20 +90,23 @@ struct HashMethodString
 protected:
     friend class columns_hashing_impl::HashMethodBase<Self, Value, Mapped, use_cache>;
 
-    static ALWAYS_INLINE void onNewKey(StringRef & key, Arena & pool)
+    static ALWAYS_INLINE void onNewKey([[maybe_unused]] StringRef & key, [[maybe_unused]] Arena & pool)
     {
-        if (key.size)
-            key.data = pool.insert(key.data, key.size);
+        if constexpr (place_string_to_arena)
+        {
+            if (key.size)
+                key.data = pool.insert(key.data, key.size);
+        }
     }
 };
 
 
 /// For the case where there is one fixed-length string key.
-template <typename Value, typename Mapped, bool use_cache = true>
+template <typename Value, typename Mapped, bool place_string_to_arena = true, bool use_cache = true>
 struct HashMethodFixedString
-    : public columns_hashing_impl::HashMethodBase<HashMethodFixedString<Value, Mapped, use_cache>, Value, Mapped, use_cache>
+    : public columns_hashing_impl::HashMethodBase<HashMethodFixedString<Value, Mapped, place_string_to_arena, use_cache>, Value, Mapped, use_cache>
 {
-    using Self = HashMethodFixedString<Value, Mapped, use_cache>;
+    using Self = HashMethodFixedString<Value, Mapped, place_string_to_arena, use_cache>;
     using Base = columns_hashing_impl::HashMethodBase<Self, Value, Mapped, use_cache>;
 
     size_t n;
@@ -123,7 +126,11 @@ struct HashMethodFixedString
 
 protected:
     friend class columns_hashing_impl::HashMethodBase<Self, Value, Mapped, use_cache>;
-    static ALWAYS_INLINE void onNewKey(StringRef & key, Arena & pool) { key.data = pool.insert(key.data, key.size); }
+    static ALWAYS_INLINE void onNewKey([[maybe_unused]] StringRef & key, [[maybe_unused]] Arena & pool)
+    {
+        if constexpr (place_string_to_arena)
+            key.data = pool.insert(key.data, key.size);
+    }
 };
 
 

dbms/src/Functions/arrayEnumerateExtended.h

@@ -73,14 +73,14 @@ private:
     {
         using Set = ClearableHashMap<StringRef, UInt32, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
                 HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
-        using Method = ColumnsHashing::HashMethodString<typename Set::value_type, UInt32, false>;
+        using Method = ColumnsHashing::HashMethodString<typename Set::value_type, UInt32, false, false>;
     };
 
     struct MethodFixedString
     {
         using Set = ClearableHashMap<StringRef, UInt32, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
                 HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
-        using Method = ColumnsHashing::HashMethodFixedString<typename Set::value_type, UInt32, false>;
+        using Method = ColumnsHashing::HashMethodFixedString<typename Set::value_type, UInt32, false, false>;
     };
 
     struct MethodFixed

dbms/src/Functions/arrayUniq.cpp

@@ -75,14 +75,14 @@ private:
     {
         using Set = ClearableHashSet<StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
                 HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
-        using Method = ColumnsHashing::HashMethodString<typename Set::value_type, void, false>;
+        using Method = ColumnsHashing::HashMethodString<typename Set::value_type, void, false, false>;
     };
 
     struct MethodFixedString
     {
         using Set = ClearableHashSet<StringRef, StringRefHash, HashTableGrower<INITIAL_SIZE_DEGREE>,
                 HashTableAllocatorWithStackMemory<(1ULL << INITIAL_SIZE_DEGREE) * sizeof(StringRef)>>;
-        using Method = ColumnsHashing::HashMethodFixedString<typename Set::value_type, void, false>;
+        using Method = ColumnsHashing::HashMethodFixedString<typename Set::value_type, void, false, false>;
     };
 
     struct MethodFixed

dbms/src/Interpreters/Join.cpp

@@ -191,11 +191,11 @@ template <typename Value, typename Mapped> struct KeyGetterForTypeImpl<Join::Typ
 };
 template <typename Value, typename Mapped> struct KeyGetterForTypeImpl<Join::Type::key_string, Value, Mapped>
 {
-    using Type = ColumnsHashing::HashMethodString<Value, Mapped, false>;
+    using Type = ColumnsHashing::HashMethodString<Value, Mapped, true, false>;
 };
 template <typename Value, typename Mapped> struct KeyGetterForTypeImpl<Join::Type::key_fixed_string, Value, Mapped>
 {
-    using Type = ColumnsHashing::HashMethodFixedString<Value, Mapped, false>;
+    using Type = ColumnsHashing::HashMethodFixedString<Value, Mapped, true, false>;
 };
 template <typename Value, typename Mapped> struct KeyGetterForTypeImpl<Join::Type::keys128, Value, Mapped>
 {

dbms/src/Interpreters/SetVariants.h

@@ -40,7 +40,7 @@ struct SetMethodString
 
     Data data;
 
-    using State = ColumnsHashing::HashMethodString<typename Data::value_type, void, false>;
+    using State = ColumnsHashing::HashMethodString<typename Data::value_type, void, true, false>;
 };
 
 /// For the case when there is one fixed-length string key.
@@ -52,7 +52,7 @@ struct SetMethodFixedString
 
     Data data;
 
-    using State = ColumnsHashing::HashMethodFixedString<typename Data::value_type, void, false>;
+    using State = ColumnsHashing::HashMethodFixedString<typename Data::value_type, void, true, false>;
 };
 
 namespace set_impl