dbms: Server: Synced with master. [#METR-17276]

dbms/include/DB/Common/CombinedCardinalityEstimator.h

@@ -203,8 +203,6 @@ private:
 		if (container_type != details::ContainerType::SMALL)
 			throw Poco::Exception("Internal error", ErrorCodes::LOGICAL_ERROR);
 
-		container_type = details::ContainerType::MEDIUM;
-
 		if (current_memory_tracker)
 			current_memory_tracker->alloc(sizeof(medium));
 
@@ -214,6 +212,8 @@ private:
 			tmp_medium->insert(x);
 
 		medium = tmp_medium;
+
+		container_type = details::ContainerType::MEDIUM;
 	}
 
 	void toLarge()
@@ -221,18 +221,26 @@ private:
 		if ((container_type != details::ContainerType::SMALL) && (container_type != details::ContainerType::MEDIUM))
 			throw Poco::Exception("Internal error", ErrorCodes::LOGICAL_ERROR);
 
-		container_type = details::ContainerType::LARGE;
-
 		if (current_memory_tracker)
 			current_memory_tracker->alloc(sizeof(large));
 
 		Large * tmp_large = new Large;
 
+		if (container_type == details::ContainerType::SMALL)
+		{
+			for (const auto & x : small)
+				tmp_large->insert(x);
+		}
+		else if (container_type == details::ContainerType::MEDIUM)
+		{
 			for (const auto & x : *medium)
 				tmp_large->insert(x);
+		}
 
 		large = tmp_large;
 
+		if (container_type == details::ContainerType::MEDIUM)
+		{
 			delete medium;
 			medium = nullptr;
 
@@ -240,6 +248,9 @@ private:
 				current_memory_tracker->free(sizeof(medium));
 		}
 
+		container_type = details::ContainerType::LARGE;
+	}
+
 private:
 	Small small;
 	Medium * medium = nullptr;

dbms/include/DB/Functions/FunctionsStringArray.h

@@ -132,7 +132,8 @@ public:
 		const ColumnConstString * col = typeid_cast<const ColumnConstString *>(&*block.getByPosition(arguments[0]).column);
 
 		if (!col)
-			throw Exception("Illegal column " + col->getName() + " of first argument of function " + getName() + ". Must be constant string.",
+			throw Exception("Illegal column " + block.getByPosition(arguments[0]).column->getName()
+				+ " of first argument of function " + getName() + ". Must be constant string.",
 				ErrorCodes::ILLEGAL_COLUMN);
 
 		const String & sep_str = col->getData();
@@ -198,7 +199,8 @@ public:
 		const ColumnConstString * col = typeid_cast<const ColumnConstString *>(&*block.getByPosition(arguments[0]).column);
 
 		if (!col)
-			throw Exception("Illegal column " + col->getName() + " of first argument of function " + getName() + ". Must be constant string.",
+			throw Exception("Illegal column " + block.getByPosition(arguments[0]).column->getName()
+				+ " of first argument of function " + getName() + ". Must be constant string.",
 				ErrorCodes::ILLEGAL_COLUMN);
 
 		sep = col->getData();
@@ -264,7 +266,8 @@ public:
 		const ColumnConstString * col = typeid_cast<const ColumnConstString *>(&*block.getByPosition(arguments[1]).column);
 
 		if (!col)
-			throw Exception("Illegal column " + col->getName() + " of first argument of function " + getName() + ". Must be constant string.",
+			throw Exception("Illegal column " + block.getByPosition(arguments[1]).column->getName()
+				+ " of first argument of function " + getName() + ". Must be constant string.",
 				ErrorCodes::ILLEGAL_COLUMN);
 
 		re = Regexps::get<false, false>(col->getData());

dbms/src/Interpreters/InterpreterSelectQuery.cpp

@@ -522,6 +522,9 @@ void InterpreterSelectQuery::executeSingleQuery()
 				if (has_order_by)
 					executeOrder(streams);
 
+				if (has_order_by && query.limit_length)
+					executeDistinct(streams, false, selected_columns);
+
 				if (query.limit_length)
 					executePreLimit(streams);
 			}

dbms/src/Interpreters/tests/hash_map_string_3.cpp

@@ -4,6 +4,9 @@
 
 #include <statdaemons/Stopwatch.h>
 
+#include <farmhash.h>
+#include <metrohash.h>
+
 #define DBMS_HASH_MAP_COUNT_COLLISIONS
 #define DBMS_HASH_MAP_DEBUG_RESIZES
 
@@ -32,8 +35,8 @@ for file in MobilePhoneModel PageCharset Params URLDomain UTMSource Referer URL
 	if [[ $TOTAL_ELEMS -gt 25000000 ]]; then break; fi
     ./hash_map_string_3 $size $method < ${file}.bin 2>&1 |
      grep HashMap | grep -oE '[0-9\.]+ elem';
-   done | awk -W interactive '{ if ($1 > x) { x = $1 }; printf(".") } END { print x }' | tee /tmp/hash_map_string_2_res;
+   done | awk -W interactive '{ if ($1 > x) { x = $1 }; printf(".") } END { print x }' | tee /tmp/hash_map_string_3_res;
-   CUR_RESULT=$(cat /tmp/hash_map_string_2_res | tr -d '.')
+   CUR_RESULT=$(cat /tmp/hash_map_string_3_res | tr -d '.')
    if [[ $CUR_RESULT -gt $BEST_RESULT ]]; then
     BEST_METHOD=$method
     BEST_RESULT=$CUR_RESULT
@@ -277,6 +280,32 @@ struct VerySimpleHash
 };
 
 
+struct FarmHash64
+{
+	size_t operator() (StringRef x) const
+	{
+		return farmhash::Hash64(x.data, x.size);
+	}
+};
+
+
+template <void metrohash64(const uint8_t * key, uint64_t len, uint32_t seed, uint8_t * out)>
+struct MetroHash64
+{
+	size_t operator() (StringRef x) const
+	{
+		union {
+			std::uint64_t u64;
+			std::uint8_t u8[sizeof(u64)];
+		};
+
+		metrohash64(reinterpret_cast<const std::uint8_t *>(x.data), x.size, 0, u8);
+
+		return u64;
+	}
+};
+
+
 /*struct CRC32Hash
 {
 	size_t operator() (StringRef x) const
@@ -426,6 +455,9 @@ int main(int argc, char ** argv)
 	if (!m || m == 5) bench<StringRef_CompareMemcmp, CRC32Hash>		(data, "StringRef_CRC32Hash");
 	if (!m || m == 6) bench<StringRef_CompareMemcmp, CRC32ILPHash>	(data, "StringRef_CRC32ILPHash");
 	if (!m || m == 7) bench<StringRef_CompareMemcmp, VerySimpleHash>(data, "StringRef_VerySimpleHash");
+	if (!m || m == 8) bench<StringRef_CompareMemcmp, FarmHash64>(data, "StringRef_FarmHash64");
+	if (!m || m == 9) bench<StringRef_CompareMemcmp, MetroHash64<metrohash64_1>>(data, "StringRef_MetroHash64_1");
+	if (!m || m == 10) bench<StringRef_CompareMemcmp, MetroHash64<metrohash64_2>>(data, "StringRef_MetroHash64_2");
 
 	return 0;
 }

dbms/tests/queries/0_stateless/00200_distinct_order_by_limit_distributed.reference

@@ -0,0 +1,10 @@
+0
+1
+2
+3
+4
+5
+6
+7
+8
+9

dbms/tests/queries/0_stateless/00200_distinct_order_by_limit_distributed.sql

@@ -0,0 +1,5 @@
+DROP TABLE IF EXISTS numbers_memory;
+CREATE TABLE numbers_memory AS system.numbers ENGINE = Memory;
+INSERT INTO numbers_memory SELECT number FROM system.numbers LIMIT 100;
+SELECT DISTINCT number FROM remote('127.0.0.{2,3}', default.numbers_memory) ORDER BY number LIMIT 10;
+DROP TABLE numbers_memory;