Merge branch 'duckdb:main' into window-progress

.github/config/extensions.csv

@@ -15,4 +15,4 @@ aws,https://github.com/duckdb/duckdb_aws,f7b8729f1cce5ada5d4add70e1486de50763fb9
 azure,https://github.com/duckdb/duckdb_azure,09623777a366572bfb8fa53e47acdf72133a360e,
 spatial,https://github.com/duckdb/duckdb_spatial,8ac803e986ccda34f32dee82a7faae95b72b3492,
 iceberg,https://github.com/duckdb/duckdb_iceberg,d89423c2ff90a0b98a093a133c8dfe2a55b9e092,
-vss,https://github.com/duckdb/duckdb_vss,9038b50cefb8bfd6b8ab8a254d3c728f3f172d15,
+vss,https://github.com/duckdb/duckdb_vss,8145f41d97178e82bed3376215eb8d02bcf1eec5,

.github/config/out_of_tree_extensions.cmake

@@ -106,6 +106,6 @@ endif()
 duckdb_extension_load(vss
         LOAD_TESTS
         GIT_URL https://github.com/duckdb/duckdb_vss
-        GIT_TAG 9038b50cefb8bfd6b8ab8a254d3c728f3f172d15
+        GIT_TAG 8145f41d97178e82bed3376215eb8d02bcf1eec5
         TEST_DIR test/sql
     )

CMakeLists.txt

@@ -83,6 +83,13 @@ if (EXTENSION_STATIC_BUILD AND "${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU")
 endif()
 
 option(DISABLE_UNITY "Disable unity builds." FALSE)
+option(USE_WASM_THREADS "Should threads be used" FALSE)
+if (${USE_WASM_THREADS})
+   set(WASM_THREAD_FLAGS
+       -pthread
+       -sSHARED_MEMORY=1
+   )
+endif()
 
 option(FORCE_COLORED_OUTPUT
        "Always produce ANSI-colored output (GNU/Clang only)." FALSE)
@@ -553,7 +560,6 @@ include_directories(third_party/fast_float)
 include_directories(third_party/re2)
 include_directories(third_party/miniz)
 include_directories(third_party/utf8proc/include)
-include_directories(third_party/miniparquet)
 include_directories(third_party/concurrentqueue)
 include_directories(third_party/pcg)
 include_directories(third_party/tdigest)
@@ -810,10 +816,11 @@ function(build_loadable_extension_directory NAME OUTPUT_DIRECTORY EXTENSION_VERS
       COMMAND ${CMAKE_COMMAND} -E copy $<TARGET_FILE:${TARGET_NAME}> $<TARGET_FILE:${TARGET_NAME}>.lib
       )
     # Compile the library into the actual wasm file
+    string(TOUPPER ${NAME} EXTENSION_NAME_UPPERCASE)
     add_custom_command(
       TARGET ${TARGET_NAME}
       POST_BUILD
-      COMMAND emcc $<TARGET_FILE:${TARGET_NAME}>.lib -o $<TARGET_FILE:${TARGET_NAME}> -sSIDE_MODULE=1 -O3
+      COMMAND emcc $<TARGET_FILE:${TARGET_NAME}>.lib -o $<TARGET_FILE:${TARGET_NAME}> -O3 -sSIDE_MODULE=2 -sEXPORTED_FUNCTIONS="_${NAME}_init" ${WASM_THREAD_FLAGS} ${DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_LINKED_LIBS}
       )
   endif()
   add_custom_command(
@@ -846,7 +853,7 @@ function(build_static_extension NAME PARAMETERS)
 endfunction()
 
 # Internal extension register function
-function(register_extension NAME DONT_LINK DONT_BUILD LOAD_TESTS PATH INCLUDE_PATH TEST_PATH)
+function(register_extension NAME DONT_LINK DONT_BUILD LOAD_TESTS PATH INCLUDE_PATH TEST_PATH LINKED_LIBS)
   string(TOLOWER ${NAME} EXTENSION_NAME_LOWERCASE)
   string(TOUPPER ${NAME} EXTENSION_NAME_UPPERCASE)
 
@@ -868,6 +875,8 @@ function(register_extension NAME DONT_LINK DONT_BUILD LOAD_TESTS PATH INCLUDE_PA
     endif()
   endif()
 
+  set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_LINKED_LIBS "${LINKED_LIBS}" PARENT_SCOPE)
+
   # Allows explicitly disabling extensions that may be specified in other configurations
   if (NOT ${DONT_BUILD} AND NOT ${EXTENSION_TESTS_ONLY})
     set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_SHOULD_BUILD TRUE PARENT_SCOPE)
@@ -893,7 +902,7 @@ function(register_extension NAME DONT_LINK DONT_BUILD LOAD_TESTS PATH INCLUDE_PA
 endfunction()
 
 # Downloads the external extension repo at the specified commit and calls register_extension
-macro(register_external_extension NAME URL COMMIT DONT_LINK DONT_BUILD LOAD_TESTS PATH INCLUDE_PATH TEST_PATH APPLY_PATCHES SUBMODULES)
+macro(register_external_extension NAME URL COMMIT DONT_LINK DONT_BUILD LOAD_TESTS PATH INCLUDE_PATH TEST_PATH APPLY_PATCHES LINKED_LIBS SUBMODULES)
   include(FetchContent)
   if (${APPLY_PATCHES})
     set(PATCH_COMMAND python3 ${CMAKE_SOURCE_DIR}/scripts/apply_extension_patches.py ${CMAKE_SOURCE_DIR}/.github/patches/extensions/${NAME}/)
@@ -921,6 +930,8 @@ macro(register_external_extension NAME URL COMMIT DONT_LINK DONT_BUILD LOAD_TEST
   endif()
 
   message(STATUS "Load extension '${NAME}' from ${URL} @ ${GIT_SHORT_COMMIT}")
+
+  string(TOUPPER ${NAME} EXTENSION_NAME_UPPERCASE)
   set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_EXT_VERSION "${GIT_SHORT_COMMIT}" PARENT_SCOPE)
 
   if ("${INCLUDE_PATH}" STREQUAL "")
@@ -935,13 +946,13 @@ macro(register_external_extension NAME URL COMMIT DONT_LINK DONT_BUILD LOAD_TEST
     set(TEST_FULL_PATH "${${NAME}_extension_fc_SOURCE_DIR}/${TEST_PATH}")
   endif()
 
-  register_extension(${NAME} ${DONT_LINK} ${DONT_BUILD} ${LOAD_TESTS} ${${NAME}_extension_fc_SOURCE_DIR}/${PATH} "${INCLUDE_FULL_PATH}" "${TEST_FULL_PATH}")
+  register_extension(${NAME} ${DONT_LINK} ${DONT_BUILD} ${LOAD_TESTS} ${${NAME}_extension_fc_SOURCE_DIR}/${PATH} "${INCLUDE_FULL_PATH}" "${TEST_FULL_PATH}" "${LINKED_LIBS}")
 endmacro()
 
 function(duckdb_extension_load NAME)
   # Parameter parsing
   set(options DONT_LINK DONT_BUILD LOAD_TESTS APPLY_PATCHES)
-  set(oneValueArgs SOURCE_DIR INCLUDE_DIR TEST_DIR GIT_URL GIT_TAG SUBMODULES EXTENSION_VERSION)
+  set(oneValueArgs SOURCE_DIR INCLUDE_DIR TEST_DIR GIT_URL GIT_TAG SUBMODULES EXTENSION_VERSION LINKED_LIBS)
   cmake_parse_arguments(duckdb_extension_load "${options}" "${oneValueArgs}" "" ${ARGN})
 
   string(TOLOWER ${NAME} EXTENSION_NAME_LOWERCASE)
@@ -960,12 +971,12 @@ function(duckdb_extension_load NAME)
 
   # Remote Git extension
   if (${duckdb_extension_load_DONT_BUILD})
-    register_extension(${NAME} "${duckdb_extension_load_DONT_LINK}" "${duckdb_extension_load_DONT_BUILD}" "" "" "" "")
+    register_extension(${NAME} "${duckdb_extension_load_DONT_LINK}" "${duckdb_extension_load_DONT_BUILD}" "" "" "" "" "")
   elseif (NOT "${duckdb_extension_load_GIT_URL}" STREQUAL "")
     if ("${duckdb_extension_load_GIT_TAG}" STREQUAL "")
       error("Git URL specified but no valid GIT_TAG was found for ${NAME} extension")
     endif()
-    register_external_extension(${NAME} "${duckdb_extension_load_GIT_URL}" "${duckdb_extension_load_GIT_TAG}" "${duckdb_extension_load_DONT_LINK}" "${duckdb_extension_load_DONT_BUILD}" "${duckdb_extension_load_LOAD_TESTS}" "${duckdb_extension_load_SOURCE_DIR}" "${duckdb_extension_load_INCLUDE_DIR}" "${duckdb_extension_load_TEST_DIR}" "${duckdb_extension_load_APPLY_PATCHES}" "${duckdb_extension_load_SUBMODULES}")
+    register_external_extension(${NAME} "${duckdb_extension_load_GIT_URL}" "${duckdb_extension_load_GIT_TAG}" "${duckdb_extension_load_DONT_LINK}" "${duckdb_extension_load_DONT_BUILD}" "${duckdb_extension_load_LOAD_TESTS}" "${duckdb_extension_load_SOURCE_DIR}" "${duckdb_extension_load_INCLUDE_DIR}" "${duckdb_extension_load_TEST_DIR}" "${duckdb_extension_load_APPLY_PATCHES}" "${duckdb_extension_load_LINKED_LIBS}" "${duckdb_extension_load_SUBMODULES}")
     if (NOT "${duckdb_extension_load_EXTENSION_VERSION}" STREQUAL "")
       set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_EXT_VERSION "${duckdb_extension_load_EXTENSION_VERSION}" PARENT_SCOPE)
     endif()
@@ -1003,7 +1014,7 @@ function(duckdb_extension_load NAME)
         set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_EXT_VERSION "" PARENT_SCOPE)
       endif()
     endif()
-    register_extension(${NAME} "${duckdb_extension_load_DONT_LINK}" "${duckdb_extension_load_DONT_BUILD}" "${duckdb_extension_load_LOAD_TESTS}" "${duckdb_extension_load_SOURCE_DIR}" "${INCLUDE_PATH_DEFAULT}" "${TEST_PATH_DEFAULT}")
+    register_extension(${NAME} "${duckdb_extension_load_DONT_LINK}" "${duckdb_extension_load_DONT_BUILD}" "${duckdb_extension_load_LOAD_TESTS}" "${duckdb_extension_load_SOURCE_DIR}" "${INCLUDE_PATH_DEFAULT}" "${TEST_PATH_DEFAULT}" "${duckdb_extension_load_LINKED_LIBS}")
   elseif(EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/extension_external/${NAME})
     # Local extension, default path
     message(STATUS "Load extension '${NAME}' from '${CMAKE_CURRENT_SOURCE_DIR}/extension_external'")
@@ -1012,7 +1023,7 @@ function(duckdb_extension_load NAME)
   else()
     # Local extension, default path
     message(STATUS "Load extension '${NAME}' from '${CMAKE_CURRENT_SOURCE_DIR}/extensions'")
-    register_extension(${NAME} ${duckdb_extension_load_DONT_LINK} "${duckdb_extension_load_DONT_BUILD}" "${duckdb_extension_load_LOAD_TESTS}" "${CMAKE_CURRENT_SOURCE_DIR}/extension/${NAME}" "${CMAKE_CURRENT_SOURCE_DIR}/extension/${NAME}/include" "${CMAKE_CURRENT_SOURCE_DIR}/extension/${NAME}/test/sql")
+    register_extension(${NAME} ${duckdb_extension_load_DONT_LINK} "${duckdb_extension_load_DONT_BUILD}" "${duckdb_extension_load_LOAD_TESTS}" "${CMAKE_CURRENT_SOURCE_DIR}/extension/${NAME}" "${CMAKE_CURRENT_SOURCE_DIR}/extension/${NAME}/include" "${CMAKE_CURRENT_SOURCE_DIR}/extension/${NAME}/test/sql" "${duckdb_extension_load_LINKED_LIBS}")
     set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_EXT_VERSION "${DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_EXT_VERSION}" PARENT_SCOPE)
   endif()
 
@@ -1024,6 +1035,7 @@ function(duckdb_extension_load NAME)
   set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_PATH ${DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_PATH} PARENT_SCOPE)
   set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_INCLUDE_PATH ${DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_INCLUDE_PATH} PARENT_SCOPE)
   set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_TEST_PATH ${DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_TEST_PATH} PARENT_SCOPE)
+  set(DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_LINKED_LIBS ${DUCKDB_EXTENSION_${EXTENSION_NAME_UPPERCASE}_LINKED_LIBS} PARENT_SCOPE)
 endfunction()
 
 if(${EXPORT_DLL_SYMBOLS})

Makefile

@@ -294,17 +294,17 @@ release: ${EXTENSION_CONFIG_STEP}
 
 wasm_mvp: ${EXTENSION_CONFIG_STEP}
 	mkdir -p ./build/wasm_mvp && \
-	emcmake cmake $(GENERATOR) ${COMMON_CMAKE_VARS} -DWASM_LOADABLE_EXTENSIONS=1 -DBUILD_EXTENSIONS_ONLY=1 -Bbuild/wasm_mvp -DCMAKE_CXX_FLAGS="-DDUCKDB_CUSTOM_PLATFORM=wasm_mvp" && \
+	emcmake cmake $(GENERATOR) -DWASM_LOADABLE_EXTENSIONS=1 -DBUILD_EXTENSIONS_ONLY=1 -Bbuild/wasm_mvp -DCMAKE_CXX_FLAGS="-DDUCKDB_CUSTOM_PLATFORM=wasm_mvp" -DDUCKDB_EXPLICIT_PLATFORM="wasm_mvp" ${COMMON_CMAKE_VARS} ${TOOLCHAIN_FLAGS} && \
 	emmake make -j8 -Cbuild/wasm_mvp
 
 wasm_eh: ${EXTENSION_CONFIG_STEP}
 	mkdir -p ./build/wasm_eh && \
-	emcmake cmake $(GENERATOR) ${COMMON_CMAKE_VARS} -DWASM_LOADABLE_EXTENSIONS=1 -DBUILD_EXTENSIONS_ONLY=1 -Bbuild/wasm_eh -DCMAKE_CXX_FLAGS="-fwasm-exceptions -DWEBDB_FAST_EXCEPTIONS=1 -DDUCKDB_CUSTOM_PLATFORM=wasm_eh" && \
+	emcmake cmake $(GENERATOR) -DWASM_LOADABLE_EXTENSIONS=1 -DBUILD_EXTENSIONS_ONLY=1 -Bbuild/wasm_eh -DCMAKE_CXX_FLAGS="-fwasm-exceptions -DWEBDB_FAST_EXCEPTIONS=1 -DDUCKDB_CUSTOM_PLATFORM=wasm_eh" -DDUCKDB_EXPLICIT_PLATFORM="wasm_eh" ${COMMON_CMAKE_VARS} ${TOOLCHAIN_FLAGS} && \
 	emmake make -j8 -Cbuild/wasm_eh
 
 wasm_threads: ${EXTENSION_CONFIG_STEP}
 	mkdir -p ./build/wasm_threads && \
-	emcmake cmake $(GENERATOR) ${COMMON_CMAKE_VARS} -DWASM_LOADABLE_EXTENSIONS=1 -DBUILD_EXTENSIONS_ONLY=1 -Bbuild/wasm_threads -DCMAKE_CXX_FLAGS="-fwasm-exceptions -DWEBDB_FAST_EXCEPTIONS=1 -DWITH_WASM_THREADS=1 -DWITH_WASM_SIMD=1 -DWITH_WASM_BULK_MEMORY=1 -DDUCKDB_CUSTOM_PLATFORM=wasm_threads" && \
+	emcmake cmake $(GENERATOR) -DWASM_LOADABLE_EXTENSIONS=1 -DBUILD_EXTENSIONS_ONLY=1 -Bbuild/wasm_threads -DCMAKE_CXX_FLAGS="-fwasm-exceptions -DWEBDB_FAST_EXCEPTIONS=1 -DWITH_WASM_THREADS=1 -DWITH_WASM_SIMD=1 -DWITH_WASM_BULK_MEMORY=1 -DDUCKDB_CUSTOM_PLATFORM=wasm_threads -pthread" -DDUCKDB_EXPLICIT_PLATFORM="wasm_threads" ${COMMON_CMAKE_VARS} -DUSE_WASM_THREADS=1 -DCMAKE_C_FLAGS="-pthread" ${TOOLCHAIN_FLAGS} && \
 	emmake make -j8 -Cbuild/wasm_threads
 
 cldebug: ${EXTENSION_CONFIG_STEP}
@@ -443,7 +443,10 @@ generate-files:
 	python3 scripts/generate_functions.py
 	python3 scripts/generate_serialization.py
 	python3 scripts/generate_enum_util.py
+	python3 tools/pythonpkg/scripts/generate_connection_code.py
 	./scripts/generate_micro_extended.sh
+# Run the formatter again after (re)generating the files
+	$(MAKE) format-main
 
 bundle-library: release
 	cd build/release && \

data/csv/decimal.csv

@@ -0,0 +1,2 @@
+col_a
+3.5215257120407011

extension/parquet/column_writer.cpp

@@ -480,7 +480,7 @@ void BasicColumnWriter::BeginWrite(ColumnWriterState &state_p) {
 		auto &page_info = state.page_info[page_idx];
 		if (page_info.row_count == 0) {
 			D_ASSERT(page_idx + 1 == state.page_info.size());
-			state.page_info.erase(state.page_info.begin() + page_idx);
+			state.page_info.erase_at(page_idx);
 			break;
 		}
 		PageWriteInformation write_info;

extension/sqlsmith/statement_simplifier.cpp

@@ -132,6 +132,8 @@ void StatementSimplifier::Simplify(unique_ptr<TableRef> &ref) {
 		SimplifyOptionalExpression(cp.condition);
 		SimplifyReplace(ref, cp.left);
 		SimplifyReplace(ref, cp.right);
+		SimplifyEnum(cp.type, JoinType::INNER);
+		SimplifyEnum(cp.ref_type, JoinRefType::REGULAR);
 		break;
 	}
 	case TableReferenceType::EXPRESSION_LIST: {

scripts/reduce_sql.py

@@ -258,7 +258,7 @@ def reduce_query_log(queries, shell, max_time_seconds=300):
     print(expected_error)
     print("===================================================")
 
-    final_query = reduce(sql_query, data_load, shell, expected_error, args.max_time)
+    final_query = reduce(sql_query, data_load, shell, expected_error, int(args.max_time))
     print("Found final reduced query")
     print("===================================================")
     print(final_query)

src/CMakeLists.txt

@@ -24,7 +24,7 @@ if("${CMAKE_CXX_COMPILER_ID}" STREQUAL "Clang" OR "${CMAKE_CXX_COMPILER_ID}"
                                                   STREQUAL "AppleClang")
   set(EXIT_TIME_DESTRUCTORS_WARNING TRUE)
   set(CMAKE_CXX_FLAGS_DEBUG
-      "${CMAKE_CXX_FLAGS_DEBUG} -Wexit-time-destructors -Wimplicit-int-conversion -Wshorten-64-to-32 -Wnarrowing"
+      "${CMAKE_CXX_FLAGS_DEBUG} -Wexit-time-destructors -Wimplicit-int-conversion -Wshorten-64-to-32 -Wnarrowing -Wsign-conversion -Wsign-compare"
   )
 endif()
 

src/common/adbc/nanoarrow/allocator.cpp

@@ -23,11 +23,11 @@
 namespace duckdb_nanoarrow {
 
 void *ArrowMalloc(int64_t size) {
-	return malloc(size);
+	return malloc(size_t(size));
 }
 
 void *ArrowRealloc(void *ptr, int64_t size) {
-	return realloc(ptr, size);
+	return realloc(ptr, size_t(size));
 }
 
 void ArrowFree(void *ptr) {

src/common/adbc/nanoarrow/metadata.cpp

@@ -78,7 +78,7 @@ int64_t ArrowMetadataSizeOf(const char *metadata) {
 
 	int64_t size = sizeof(int32_t);
 	while (ArrowMetadataReaderRead(&reader, &key, &value) == NANOARROW_OK) {
-		size += sizeof(int32_t) + key.n_bytes + sizeof(int32_t) + value.n_bytes;
+		size += sizeof(int32_t) + uint64_t(key.n_bytes) + sizeof(int32_t) + uint64_t(value.n_bytes);
 	}
 
 	return size;
@@ -89,7 +89,7 @@ ArrowErrorCode ArrowMetadataGetValue(const char *metadata, const char *key, cons
 	struct ArrowStringView target_key_view = {key, static_cast<int64_t>(strlen(key))};
 	value_out->data = default_value;
 	if (default_value != NULL) {
-		value_out->n_bytes = strlen(default_value);
+		value_out->n_bytes = int64_t(strlen(default_value));
 	} else {
 		value_out->n_bytes = 0;
 	}
@@ -101,7 +101,7 @@ ArrowErrorCode ArrowMetadataGetValue(const char *metadata, const char *key, cons
 
 	while (ArrowMetadataReaderRead(&reader, &key_view, &value) == NANOARROW_OK) {
 		int key_equal = target_key_view.n_bytes == key_view.n_bytes &&
-		                strncmp(target_key_view.data, key_view.data, key_view.n_bytes) == 0;
+		                strncmp(target_key_view.data, key_view.data, size_t(key_view.n_bytes)) == 0;
 		if (key_equal) {
 			value_out->data = value.data;
 			value_out->n_bytes = value.n_bytes;

src/common/adbc/nanoarrow/schema.cpp

@@ -318,7 +318,7 @@ ArrowErrorCode ArrowSchemaSetFormat(struct ArrowSchema *schema, const char *form
 
 	if (format != NULL) {
 		size_t format_size = strlen(format) + 1;
-		schema->format = (const char *)ArrowMalloc(format_size);
+		schema->format = (const char *)ArrowMalloc(int64_t(format_size));
 		if (schema->format == NULL) {
 			return ENOMEM;
 		}
@@ -338,7 +338,7 @@ ArrowErrorCode ArrowSchemaSetName(struct ArrowSchema *schema, const char *name)
 
 	if (name != NULL) {
 		size_t name_size = strlen(name) + 1;
-		schema->name = (const char *)ArrowMalloc(name_size);
+		schema->name = (const char *)ArrowMalloc(int64_t(name_size));
 		if (schema->name == NULL) {
 			return ENOMEM;
 		}
@@ -357,13 +357,13 @@ ArrowErrorCode ArrowSchemaSetMetadata(struct ArrowSchema *schema, const char *me
 	}
 
 	if (metadata != NULL) {
-		size_t metadata_size = ArrowMetadataSizeOf(metadata);
+		auto metadata_size = ArrowMetadataSizeOf(metadata);
 		schema->metadata = (const char *)ArrowMalloc(metadata_size);
 		if (schema->metadata == NULL) {
 			return ENOMEM;
 		}
 
-		memcpy((void *)schema->metadata, metadata, metadata_size);
+		memcpy((void *)schema->metadata, metadata, size_t(metadata_size));
 	} else {
 		schema->metadata = NULL;
 	}
@@ -377,15 +377,16 @@ ArrowErrorCode ArrowSchemaAllocateChildren(struct ArrowSchema *schema, int64_t n
 	}
 
 	if (n_children > 0) {
-		schema->children = (struct ArrowSchema **)ArrowMalloc(n_children * sizeof(struct ArrowSchema *));
+		schema->children =
+		    (struct ArrowSchema **)ArrowMalloc(int64_t(uint64_t(n_children) * sizeof(struct ArrowSchema *)));
 
 		if (schema->children == NULL) {
 			return ENOMEM;
 		}
 
 		schema->n_children = n_children;
 
-		memset(schema->children, 0, n_children * sizeof(struct ArrowSchema *));
+		memset(schema->children, 0, uint64_t(n_children) * sizeof(struct ArrowSchema *));
 
 		for (int64_t i = 0; i < n_children; i++) {
 			schema->children[i] = (struct ArrowSchema *)ArrowMalloc(sizeof(struct ArrowSchema));

src/common/arrow/appender/bool_data.cpp

@@ -6,6 +6,7 @@ namespace duckdb {
 void ArrowBoolData::Initialize(ArrowAppendData &result, const LogicalType &type, idx_t capacity) {
 	auto byte_count = (capacity + 7) / 8;
 	result.main_buffer.reserve(byte_count);
+	(void)AppendValidity; // silence a compiler warning about unused static function
 }
 
 void ArrowBoolData::Append(ArrowAppendData &append_data, Vector &input, idx_t from, idx_t to, idx_t input_size) {

src/common/arrow/appender/struct_data.cpp

@@ -35,7 +35,7 @@ void ArrowStructData::Finalize(ArrowAppendData &append_data, const LogicalType &
 	auto &child_types = StructType::GetChildTypes(type);
 	ArrowAppender::AddChildren(append_data, child_types.size());
 	result->children = append_data.child_pointers.data();
-	result->n_children = child_types.size();
+	result->n_children = NumericCast<int64_t>(child_types.size());
 	for (idx_t i = 0; i < child_types.size(); i++) {
 		auto &child_type = child_types[i].second;
 		append_data.child_arrays[i] = *ArrowAppender::FinalizeChild(child_type, std::move(append_data.child_data[i]));

src/common/arrow/appender/union_data.cpp

@@ -14,6 +14,7 @@ void ArrowUnionData::Initialize(ArrowAppendData &result, const LogicalType &type
 		auto child_buffer = ArrowAppender::InitializeChild(child.second, capacity, result.options);
 		result.child_data.push_back(std::move(child_buffer));
 	}
+	(void)AppendValidity; // silence a compiler warning about unused static functiondep
 }
 
 void ArrowUnionData::Append(ArrowAppendData &append_data, Vector &input, idx_t from, idx_t to, idx_t input_size) {
@@ -61,7 +62,7 @@ void ArrowUnionData::Finalize(ArrowAppendData &append_data, const LogicalType &t
 	auto &child_types = UnionType::CopyMemberTypes(type);
 	ArrowAppender::AddChildren(append_data, child_types.size());
 	result->children = append_data.child_pointers.data();
-	result->n_children = child_types.size();
+	result->n_children = NumericCast<int64_t>(child_types.size());
 	for (idx_t i = 0; i < child_types.size(); i++) {
 		auto &child_type = child_types[i].second;
 		append_data.child_arrays[i] = *ArrowAppender::FinalizeChild(child_type, std::move(append_data.child_data[i]));