Merge d78bbae17b into 682eb4c92f

SettingsChangesHistory: disable allow_experimental_shared_set_join

docs/en/engines/database-engines/atomic.md

@@ -6,7 +6,7 @@ sidebar_position: 10
 
 # Atomic 
 
-It supports non-blocking [DROP TABLE](#drop-detach-table) and [RENAME TABLE](#rename-table) queries and atomic [EXCHANGE TABLES](#exchange-tables) queries. `Atomic` database engine is used by default.
+It supports non-blocking [DROP TABLE](#drop-detach-table) and [RENAME TABLE](#rename-table) queries and atomic [EXCHANGE TABLES](#exchange-tables) queries. `Atomic` database engine is used by default. Note that on ClickHouse Cloud, the `Replicated` database engine is used by default.
 
 ## Creating a Database {#creating-a-database}
 

src/Analyzer/Passes/FunctionToSubcolumnsPass.cpp

@@ -46,7 +46,7 @@ using NodeToSubcolumnTransformer = std::function<void(QueryTreeNodePtr &, Functi
 
 void optimizeFunctionLength(QueryTreeNodePtr & node, FunctionNode &, ColumnContext & ctx)
 {
-    /// Replace `length(argument)` with `argument.size0`
+    /// Replace `length(argument)` with `argument.size0`.
     /// `argument` may be Array or Map.
 
     NameAndTypePair column{ctx.column.name + ".size0", std::make_shared<DataTypeUInt64>()};
@@ -56,8 +56,8 @@ void optimizeFunctionLength(QueryTreeNodePtr & node, FunctionNode &, ColumnConte
 template <bool positive>
 void optimizeFunctionEmpty(QueryTreeNodePtr &, FunctionNode & function_node, ColumnContext & ctx)
 {
-    /// Replace `empty(argument)` with `equals(argument.size0, 0)` if positive
-    /// Replace `notEmpty(argument)` with `notEquals(argument.size0, 0)` if not positive
+    /// Replace `empty(argument)` with `equals(argument.size0, 0)` if positive.
+    /// Replace `notEmpty(argument)` with `notEquals(argument.size0, 0)` if not positive.
     /// `argument` may be Array or Map.
 
     NameAndTypePair column{ctx.column.name + ".size0", std::make_shared<DataTypeUInt64>()};
@@ -136,19 +136,25 @@ std::map<std::pair<TypeIndex, String>, NodeToSubcolumnTransformer> node_transfor
     },
     {
         {TypeIndex::Map, "mapKeys"},
-        [](QueryTreeNodePtr & node, FunctionNode & function_node, ColumnContext & ctx)
+        [](QueryTreeNodePtr & node, FunctionNode &, ColumnContext & ctx)
         {
             /// Replace `mapKeys(map_argument)` with `map_argument.keys`
-            NameAndTypePair column{ctx.column.name + ".keys", function_node.getResultType()};
+            const auto & data_type_map = assert_cast<const DataTypeMap &>(*ctx.column.type);
+            auto key_type = std::make_shared<DataTypeArray>(data_type_map.getKeyType());
+
+            NameAndTypePair column{ctx.column.name + ".keys", key_type};
             node = std::make_shared<ColumnNode>(column, ctx.column_source);
         },
     },
     {
         {TypeIndex::Map, "mapValues"},
-        [](QueryTreeNodePtr & node, FunctionNode & function_node, ColumnContext & ctx)
+        [](QueryTreeNodePtr & node, FunctionNode &, ColumnContext & ctx)
         {
             /// Replace `mapValues(map_argument)` with `map_argument.values`
-            NameAndTypePair column{ctx.column.name + ".values", function_node.getResultType()};
+            const auto & data_type_map = assert_cast<const DataTypeMap &>(*ctx.column.type);
+            auto value_type = std::make_shared<DataTypeArray>(data_type_map.getValueType());
+
+            NameAndTypePair column{ctx.column.name + ".values", value_type};
             node = std::make_shared<ColumnNode>(column, ctx.column_source);
         },
     },
@@ -439,11 +445,16 @@ public:
         if (!identifiers_to_optimize.contains(qualified_name))
             return;
 
+        auto result_type = function_node->getResultType();
         auto transformer_it = node_transformers.find({column.type->getTypeId(), function_node->getFunctionName()});
+
         if (transformer_it != node_transformers.end())
         {
             ColumnContext ctx{std::move(column), first_argument_column_node->getColumnSource(), getContext()};
             transformer_it->second(node, *function_node, ctx);
+
+            if (!result_type->equals(*node->getResultType()))
+                node = buildCastFunction(node, result_type, getContext());
         }
     }
 };

src/Core/SettingsChangesHistory.cpp

@@ -117,7 +117,7 @@ static std::initializer_list<std::pair<ClickHouseVersion, SettingsChangesHistory
             {"min_free_disk_ratio_to_perform_insert", 0.0, 0.0, "New setting."},
             {"enable_named_columns_in_function_tuple", false, false, "Disabled pending usability improvements"},
             {"cloud_mode_database_engine", 1, 1, "A setting for ClickHouse Cloud"},
-            {"allow_experimental_shared_set_join", 1, 1, "A setting for ClickHouse Cloud"},
+            {"allow_experimental_shared_set_join", 0, 0, "A setting for ClickHouse Cloud"},
             {"read_through_distributed_cache", 0, 0, "A setting for ClickHouse Cloud"},
             {"write_through_distributed_cache", 0, 0, "A setting for ClickHouse Cloud"},
             {"distributed_cache_throw_on_error", 0, 0, "A setting for ClickHouse Cloud"},

tests/queries/0_stateless/03233_insert_into_distributed_table_bug.sh

@@ -0,0 +1,51 @@
+#!/usr/bin/env bash
+# Tags: distributed
+# https://github.com/ClickHouse/ClickHouse/issues/65520
+
+CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+# shellcheck source=../shell_config.sh
+. "$CURDIR"/../shell_config.sh
+
+
+${CLICKHOUSE_CLIENT} -q "drop database if exists ${CLICKHOUSE_DATABASE}_db on cluster test_shard_localhost"
+${CLICKHOUSE_CLIENT} -q "create database ${CLICKHOUSE_DATABASE}_db on cluster test_shard_localhost"
+
+${CLICKHOUSE_CLIENT} -q "drop table if exists ${CLICKHOUSE_DATABASE}_db.segfault_table on cluster test_shard_localhost"
+${CLICKHOUSE_CLIENT} -q "drop table if exists ${CLICKHOUSE_DATABASE}_db.__segfault_table on cluster test_shard_localhost"
+
+
+${CLICKHOUSE_CLIENT} -q "
+create table ${CLICKHOUSE_DATABASE}_db.__segfault_table on cluster test_shard_localhost ( 
+  c_mpcnr33 Int32 primary key,
+  c_v8s String,
+  c_l Int32,
+  c_jismi1 String,
+  c_p37t64z75 Bool not null,
+  c_uz Bool,
+  c_rp Int32 primary key,
+  c_d56dwp13jp Bool,
+  c_sf__xnd4 Float64 not null,
+)"
+
+${CLICKHOUSE_CLIENT} -q "create table ${CLICKHOUSE_DATABASE}_db.segfault_table on cluster test_shard_localhost as ${CLICKHOUSE_DATABASE}_db.__segfault_table ENGINE = Distributed(test_shard_localhost, ${CLICKHOUSE_DATABASE}_db, __segfault_table, c_mpcnr33)"
+
+${CLICKHOUSE_CLIENT} -q "
+create table ${CLICKHOUSE_DATABASE}_db.__t_nh1w on cluster test_shard_localhost ( 
+  c_sfdzg Int32,
+  c_xf Bool,
+  c_u3xs92nr4c String,
+  c_b_m Int32 primary key,
+  c_lgy Int32,
+)"
+
+${CLICKHOUSE_CLIENT} -q "create table ${CLICKHOUSE_DATABASE}_db.t_nh1w on cluster test_shard_localhost as ${CLICKHOUSE_DATABASE}_db.__t_nh1w ENGINE = Distributed(test_shard_localhost, ${CLICKHOUSE_DATABASE}_db, __t_nh1w, c_b_m)"
+
+
+query="insert into ${CLICKHOUSE_DATABASE}_db.segfault_table (c_mpcnr33, c_v8s, c_l, c_jismi1, c_p37t64z75, c_uz, c_rp, c_d56dwp13jp, c_sf__xnd4) values (868701807, coalesce((select c_u3xs92nr4c from ${CLICKHOUSE_DATABASE}_db.t_nh1w order by c_u3xs92nr4c limit 1 offset 6), 'llwlzwb3'), 1824351772, coalesce(MACNumToString(lcm(-3, -6)), 'f'))"
+
+# This was triggering a crash in the previous versions. The query can produce an error but should not lead to a crash.
+curl -d@- -sS "${CLICKHOUSE_URL}" <<< "$query" >/dev/null 2>&1 ||:
+
+
+${CLICKHOUSE_CLIENT} -q "drop table if exists ${CLICKHOUSE_DATABASE}_db.segfault_table on cluster test_shard_localhost"
+${CLICKHOUSE_CLIENT} -q "drop table if exists ${CLICKHOUSE_DATABASE}_db.__segfault_table on cluster test_shard_localhost"

tests/queries/0_stateless/03276_functions_to_subcolumns_lc.reference

@@ -0,0 +1 @@
+['foo']	['bar']

tests/queries/0_stateless/03276_functions_to_subcolumns_lc.sql

@@ -0,0 +1,14 @@
+DROP TABLE IF EXISTS t_map_lc;
+
+CREATE TABLE t_map_lc
+(
+    kv Map(LowCardinality(String), LowCardinality(String)),
+    k Array(LowCardinality(String)) ALIAS mapKeys(kv),
+    v Array(LowCardinality(String)) ALIAS mapValues(kv)
+) ENGINE = Memory;
+
+INSERT INTO t_map_lc VALUES (map('foo', 'bar'));
+
+SELECT k, v FROM t_map_lc SETTINGS optimize_functions_to_subcolumns=1;
+
+DROP TABLE t_map_lc;