Merge branch 'master' into fts_attach

.github/config/uncovered_files.csv

@@ -69,7 +69,7 @@ common/types/blob.cpp	3
 common/types/chunk_collection.cpp	94
 common/types/column/column_data_allocator.cpp	13
 common/types/column/column_data_collection.cpp	50
-common/types/column/partitioned_column_data.cpp	6
+common/types/column/partitioned_column_data.cpp	8
 common/types/conflict_info.cpp	2
 common/types/conflict_manager.cpp	3
 common/types/data_chunk.cpp	19
@@ -342,7 +342,7 @@ include/duckdb/common/operator/numeric_cast.hpp	2
 include/duckdb/common/operator/subtract.hpp	2
 include/duckdb/common/pipe_file_system.hpp	3
 include/duckdb/common/radix.hpp	2
-include/duckdb/common/radix_partitioning.hpp	3
+include/duckdb/common/radix_partitioning.hpp	5
 include/duckdb/common/re2_regex.hpp	21
 include/duckdb/common/serializer.hpp	4
 include/duckdb/common/serializer/format_deserializer.hpp	37

.github/workflows/NightlyTests.yml

@@ -26,7 +26,7 @@ concurrency:
 
 env:
   GH_TOKEN: ${{ secrets.GH_TOKEN }}
-  DUCKDB_WASM_VERSION: "0e27318"
+  DUCKDB_WASM_VERSION: "a8f2c38"
   CCACHE_SAVE: ${{ github.repository != 'duckdb/duckdb' }}
 
 jobs:
@@ -688,7 +688,7 @@ jobs:
 
     - uses: mymindstorm/setup-emsdk@v12
       with:
-        version: 3.1.41
+        version: 'latest'
 
     - name: Setup
       shell: bash

CMakeLists.txt

@@ -692,6 +692,14 @@ function(build_loadable_extension NAME PARAMETERS)
   build_loadable_extension_directory(${NAME} "extension/${NAME}" "${PARAMETERS}" ${FILES})
 endfunction()
 
+function(build_static_extension NAME PARAMETERS)
+  # all parameters after name
+  set(FILES ${ARGV})
+  list(REMOVE_AT FILES 0)
+  add_library(${NAME}_extension STATIC ${FILES})
+  target_link_libraries(${NAME}_extension duckdb_static)
+endfunction()
+
 # Internal extension register function
 function(register_extension NAME DONT_LINK DONT_BUILD LOAD_TESTS PATH INCLUDE_PATH TEST_PATH)
   string(TOLOWER ${NAME} EXTENSION_NAME_LOWERCASE)

extension/autocomplete/CMakeLists.txt

@@ -6,7 +6,7 @@ include_directories(include)
 
 set(AUTOCOMPLETE_EXTENSION_FILES autocomplete_extension.cpp)
 
-add_library(autocomplete_extension STATIC ${AUTOCOMPLETE_EXTENSION_FILES})
+build_static_extension(autocomplete ${AUTOCOMPLETE_EXTENSION_FILES})
 set(PARAMETERS "-warnings")
 build_loadable_extension(autocomplete ${PARAMETERS}
                          ${AUTOCOMPLETE_EXTENSION_FILES})

extension/excel/CMakeLists.txt

@@ -6,8 +6,7 @@ include_directories(numformat/include)
 include_directories(include)
 add_subdirectory(numformat)
 
-add_library(excel_extension STATIC excel_extension.cpp
-                                   ${NUMFORMAT_OBJECT_FILES})
+build_static_extension(excel excel_extension.cpp ${NUMFORMAT_OBJECT_FILES})
 set(PARAMETERS "-warnings")
 build_loadable_extension(excel ${PARAMETERS} excel_extension.cpp
                          ${NUMFORMAT_OBJECT_FILES})

extension/fts/CMakeLists.txt

@@ -40,7 +40,7 @@ set(FTS_SOURCES
     ../../third_party/snowball/src_c/stem_UTF_8_tamil.cpp
     ../../third_party/snowball/src_c/stem_UTF_8_turkish.cpp)
 
-add_library(fts_extension STATIC ${FTS_SOURCES})
+build_static_extension(fts ${FTS_SOURCES})
 set(PARAMETERS "-warnings")
 build_loadable_extension(fts ${PARAMETERS} ${FTS_SOURCES})
 

extension/httpfs/CMakeLists.txt

@@ -6,8 +6,8 @@ add_extension_definitions()
 
 include_directories(include ../../third_party/httplib ../parquet/include)
 
-add_library(httpfs_extension STATIC s3fs.cpp httpfs.cpp crypto.cpp
-                                    httpfs_extension.cpp)
+build_static_extension(httpfs s3fs.cpp httpfs.cpp crypto.cpp
+                       httpfs_extension.cpp)
 set(PARAMETERS "-warnings")
 build_loadable_extension(httpfs ${PARAMETERS} s3fs.cpp httpfs.cpp crypto.cpp
                          httpfs_extension.cpp)

extension/icu/CMakeLists.txt

@@ -23,7 +23,7 @@ set(ICU_EXTENSION_FILES
     icu-timebucket.cpp
     icu-timezone.cpp)
 
-add_library(icu_extension STATIC ${ICU_EXTENSION_FILES})
+build_static_extension(icu ${ICU_EXTENSION_FILES})
 link_threads(icu_extension)
 disable_target_warnings(icu_extension)
 set(PARAMETERS "-no-warnings")

extension/jemalloc/CMakeLists.txt

@@ -9,8 +9,7 @@ add_subdirectory(jemalloc)
 
 set(JEMALLOC_EXTENSION_FILES jemalloc_extension.cpp ${JEMALLOC_OBJECT_FILES})
 
-add_library(jemalloc_extension STATIC ${JEMALLOC_EXTENSION_FILES})
-
+build_static_extension(jemalloc ${JEMALLOC_EXTENSION_FILES})
 # we do not do build_loadable_extension here because jemalloc is static-only
 
 install(

extension/json/CMakeLists.txt

@@ -31,7 +31,7 @@ set(JSON_EXTENSION_FILES
     json_functions/read_json_objects.cpp
     ${YYJSON_OBJECT_FILES})
 
-add_library(json_extension STATIC ${JSON_EXTENSION_FILES})
+build_static_extension(json ${JSON_EXTENSION_FILES})
 set(PARAMETERS "-warnings")
 build_loadable_extension(json ${PARAMETERS} ${JSON_EXTENSION_FILES})
 

extension/parquet/CMakeLists.txt

@@ -51,7 +51,8 @@ if(NOT CLANG_TIDY)
       ../../third_party/zstd/compress/zstd_opt.cpp)
 endif()
 
-add_library(parquet_extension STATIC ${PARQUET_EXTENSION_FILES})
+build_static_extension(parquet ${PARQUET_EXTENSION_FILES})
+
 set(PARAMETERS "-warnings")
 build_loadable_extension(parquet ${PARAMETERS} ${PARQUET_EXTENSION_FILES})
 

extension/sqlsmith/CMakeLists.txt

@@ -10,7 +10,7 @@ set(SQLSMITH_SOURCES
     sqlsmith_extension.cpp statement_generator.cpp statement_simplifier.cpp
     fuzzyduck.cpp ${SQLSMITH_OBJECT_FILES})
 
-add_library(sqlsmith_extension STATIC ${SQLSMITH_SOURCES})
+build_static_extension(sqlsmith ${SQLSMITH_SOURCES})
 set(PARAMETERS "-warnings")
 build_loadable_extension(sqlsmith ${PARAMETERS} ${SQLSMITH_SOURCES})
 

extension/tpcds/CMakeLists.txt

@@ -6,7 +6,7 @@ include_directories(include)
 include_directories(dsdgen/include)
 add_subdirectory(dsdgen)
 
-add_library(tpcds_extension STATIC tpcds_extension.cpp ${DSDGEN_OBJECT_FILES})
+build_static_extension(tpcds tpcds_extension.cpp ${DSDGEN_OBJECT_FILES})
 set(PARAMETERS "-warnings")
 build_loadable_extension(tpcds ${PARAMETERS} tpcds_extension.cpp
                          ${DSDGEN_OBJECT_FILES})

extension/tpch/CMakeLists.txt

@@ -6,8 +6,7 @@ include_directories(dbgen/include)
 include_directories(include)
 add_subdirectory(dbgen)
 
-add_library(tpch_extension STATIC tpch_extension.cpp ${DBGEN_OBJECT_FILES})
-
+build_static_extension(tpch tpch_extension.cpp ${DBGEN_OBJECT_FILES})
 set(PARAMETERS "-warnings")
 build_loadable_extension(tpch ${PARAMETERS} tpch_extension.cpp
                          ${DBGEN_OBJECT_FILES})

extension/visualizer/CMakeLists.txt

@@ -4,7 +4,7 @@ project(VisualizerExtension)
 
 include_directories(include)
 
-add_library(visualizer_extension STATIC visualizer_extension.cpp)
+build_static_extension(visualizer visualizer_extension.cpp)
 set(PARAMETERS "-warnings")
 build_loadable_extension(visualizer ${PARAMETERS} visualizer_extension.cpp)
 install(

src/common/sort/partition_state.cpp

@@ -87,16 +87,22 @@ PartitionGlobalSinkState::PartitionGlobalSinkState(ClientContext &context,
                                                    const vector<unique_ptr<BaseStatistics>> &partition_stats,
                                                    idx_t estimated_cardinality)
     : context(context), buffer_manager(BufferManager::GetBufferManager(context)), allocator(Allocator::Get(context)),
-      fixed_bits(0), payload_types(payload_types), memory_per_thread(0), count(0) {
+      fixed_bits(0), payload_types(payload_types), memory_per_thread(0), max_bits(1), count(0) {
 
 	GenerateOrderings(partitions, orders, partition_bys, order_bys, partition_stats);
 
 	memory_per_thread = PhysicalOperator::GetMaxThreadMemory(context);
 	external = ClientConfig::GetConfig(context).force_external;
 
+	const auto thread_pages = PreviousPowerOfTwo(memory_per_thread / (4 * idx_t(Storage::BLOCK_ALLOC_SIZE)));
+	while (max_bits < 10 && (thread_pages >> max_bits) > 1) {
+		++max_bits;
+	}
+
 	if (!orders.empty()) {
-		grouping_types = payload_types;
-		grouping_types.push_back(LogicalType::HASH);
+		auto types = payload_types;
+		types.push_back(LogicalType::HASH);
+		grouping_types.Initialize(types);
 
 		ResizeGroupingData(estimated_cardinality);
 	}
@@ -108,10 +114,15 @@ void PartitionGlobalSinkState::SyncPartitioning(const PartitionGlobalSinkState &
 	const auto old_bits = grouping_data ? grouping_data->GetRadixBits() : 0;
 	if (fixed_bits != old_bits) {
 		const auto hash_col_idx = payload_types.size();
-		grouping_data = make_uniq<RadixPartitionedColumnData>(context, grouping_types, fixed_bits, hash_col_idx);
+		grouping_data = make_uniq<RadixPartitionedTupleData>(buffer_manager, grouping_types, fixed_bits, hash_col_idx);
 	}
 }
 
+unique_ptr<RadixPartitionedTupleData> PartitionGlobalSinkState::CreatePartition(idx_t new_bits) const {
+	const auto hash_col_idx = payload_types.size();
+	return make_uniq<RadixPartitionedTupleData>(buffer_manager, grouping_types, new_bits, hash_col_idx);
+}
+
 void PartitionGlobalSinkState::ResizeGroupingData(idx_t cardinality) {
 	//	Have we started to combine? Then just live with it.
 	if (fixed_bits || (grouping_data && !grouping_data->GetPartitions().empty())) {
@@ -121,47 +132,31 @@ void PartitionGlobalSinkState::ResizeGroupingData(idx_t cardinality) {
 	const idx_t partition_size = STANDARD_ROW_GROUPS_SIZE;
 	const auto bits = grouping_data ? grouping_data->GetRadixBits() : 0;
 	auto new_bits = bits ? bits : 4;
-	while (new_bits < 10 && (cardinality / RadixPartitioning::NumberOfPartitions(new_bits)) > partition_size) {
+	while (new_bits < max_bits && (cardinality / RadixPartitioning::NumberOfPartitions(new_bits)) > partition_size) {
 		++new_bits;
 	}
 
 	// Repartition the grouping data
 	if (new_bits != bits) {
-		const auto hash_col_idx = payload_types.size();
-		grouping_data = make_uniq<RadixPartitionedColumnData>(context, grouping_types, new_bits, hash_col_idx);
+		grouping_data = CreatePartition(new_bits);
 	}
 }
 
 void PartitionGlobalSinkState::SyncLocalPartition(GroupingPartition &local_partition, GroupingAppend &local_append) {
 	// We are done if the local_partition is right sized.
-	auto &local_radix = local_partition->Cast<RadixPartitionedColumnData>();
-	if (local_radix.GetRadixBits() == grouping_data->GetRadixBits()) {
+	auto &local_radix = local_partition->Cast<RadixPartitionedTupleData>();
+	const auto new_bits = grouping_data->GetRadixBits();
+	if (local_radix.GetRadixBits() == new_bits) {
 		return;
 	}
 
 	// If the local partition is now too small, flush it and reallocate
-	auto new_partition = grouping_data->CreateShared();
-	auto new_append = make_uniq<PartitionedColumnDataAppendState>();
-	new_partition->InitializeAppendState(*new_append);
-
+	auto new_partition = CreatePartition(new_bits);
 	local_partition->FlushAppendState(*local_append);
-	auto &local_groups = local_partition->GetPartitions();
-	for (auto &local_group : local_groups) {
-		ColumnDataScanState scanner;
-		local_group->InitializeScan(scanner);
-
-		DataChunk scan_chunk;
-		local_group->InitializeScanChunk(scan_chunk);
-		for (scan_chunk.Reset(); local_group->Scan(scanner, scan_chunk); scan_chunk.Reset()) {
-			new_partition->Append(*new_append, scan_chunk);
-		}
-	}
-
-	// The append state has stale pointers to the old local partition, so nuke it from orbit.
-	new_partition->FlushAppendState(*new_append);
+	local_partition->Repartition(*new_partition);
 
 	local_partition = std::move(new_partition);
-	local_append = make_uniq<PartitionedColumnDataAppendState>();
+	local_append = make_uniq<PartitionedTupleDataAppendState>();
 	local_partition->InitializeAppendState(*local_append);
 }
 
@@ -170,8 +165,8 @@ void PartitionGlobalSinkState::UpdateLocalPartition(GroupingPartition &local_par
 	lock_guard<mutex> guard(lock);
 
 	if (!local_partition) {
-		local_partition = grouping_data->CreateShared();
-		local_append = make_uniq<PartitionedColumnDataAppendState>();
+		local_partition = CreatePartition(grouping_data->GetRadixBits());
+		local_append = make_uniq<PartitionedTupleDataAppendState>();
 		local_partition->InitializeAppendState(*local_append);
 		return;
 	}
@@ -196,7 +191,7 @@ void PartitionGlobalSinkState::CombineLocalPartition(GroupingPartition &local_pa
 	grouping_data->Combine(*local_partition);
 }
 
-void PartitionGlobalSinkState::BuildSortState(ColumnDataCollection &group_data, GlobalSortState &global_sort) const {
+void PartitionGlobalSinkState::BuildSortState(TupleDataCollection &group_data, GlobalSortState &global_sort) const {
 	//	 Set up the sort expression computation.
 	vector<LogicalType> sort_types;
 	ExpressionExecutor executor(context);
@@ -221,30 +216,22 @@ void PartitionGlobalSinkState::BuildSortState(ColumnDataCollection &group_data,
 	for (column_t i = 0; i < payload_types.size(); ++i) {
 		column_ids.emplace_back(i);
 	}
-	ColumnDataConsumer scanner(group_data, column_ids);
-	ColumnDataConsumerScanState chunk_state;
-	chunk_state.current_chunk_state.properties = ColumnDataScanProperties::ALLOW_ZERO_COPY;
-	scanner.InitializeScan();
-	for (auto chunk_idx = scanner.ChunkCount(); chunk_idx-- > 0;) {
-		if (!scanner.AssignChunk(chunk_state)) {
-			break;
-		}
-		scanner.ScanChunk(chunk_state, payload_chunk);
-
+	TupleDataScanState chunk_state;
+	group_data.InitializeScan(chunk_state, column_ids);
+	while (group_data.Scan(chunk_state, payload_chunk)) {
 		sort_chunk.Reset();
 		executor.Execute(payload_chunk, sort_chunk);
 
 		local_sort.SinkChunk(sort_chunk, payload_chunk);
 		if (local_sort.SizeInBytes() > memory_per_thread) {
 			local_sort.Sort(global_sort, true);
 		}
-		scanner.FinishChunk(chunk_state);
 	}
 
 	global_sort.AddLocalState(local_sort);
 }
 
-void PartitionGlobalSinkState::BuildSortState(ColumnDataCollection &group_data, PartitionGlobalHashGroup &hash_group) {
+void PartitionGlobalSinkState::BuildSortState(TupleDataCollection &group_data, PartitionGlobalHashGroup &hash_group) {
 	BuildSortState(group_data, *hash_group.global_sort);
 
 	hash_group.count += group_data.Count();

src/include/duckdb/common/opener_file_system.hpp

@@ -99,11 +99,11 @@ class OpenerFileSystem : public FileSystem {
 	bool IsPipe(const string &filename) override {
 		return GetFileSystem().IsPipe(filename);
 	}
-	virtual void RemoveFile(const string &filename) override {
+	void RemoveFile(const string &filename) override {
 		GetFileSystem().RemoveFile(filename);
 	}
 
-	virtual vector<string> Glob(const string &path, FileOpener *opener = nullptr) override {
+	vector<string> Glob(const string &path, FileOpener *opener = nullptr) override {
 		if (opener) {
 			throw InternalException("OpenerFileSystem cannot take an opener - the opener is pushed automatically");
 		}