Skip to content
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions .github/workflows/cpp.yml
Original file line number Diff line number Diff line change
Expand Up @@ -290,7 +290,7 @@ jobs:
include:
- os: windows-2022
simd-level: AVX2
title: AMD64 Windows 2022 AVX2
title: AMD64 Windows 2022 C++20 AVX2
env:
ARROW_BOOST_USE_SHARED: OFF
ARROW_BUILD_BENCHMARKS: ON
Expand All @@ -317,7 +317,7 @@ jobs:
ARROW_WITH_ZLIB: ON
ARROW_WITH_ZSTD: ON
BOOST_SOURCE: BUNDLED
CMAKE_CXX_STANDARD: "17"
CMAKE_CXX_STANDARD: "20"
CMAKE_GENERATOR: Ninja
CMAKE_INSTALL_LIBDIR: bin
CMAKE_INSTALL_PREFIX: /usr
Expand Down
3 changes: 2 additions & 1 deletion .github/workflows/r.yml
Original file line number Diff line number Diff line change
Expand Up @@ -286,12 +286,13 @@ jobs:
# static library. The R is not used here but R 4.1 was the last R to use
# Rtools40.
r-version: "4.1"
rtools-version: 40
rtools-version: ${{ matrix.config.rtools }}
Ncpus: 2
- name: Build Arrow C++
shell: bash
env:
MINGW_ARCH: ${{ matrix.config.arch }}
RTOOLS_VERSION: ${{ matrix.config.rtools }}
run: ci/scripts/r_windows_build.sh
- name: Rename libarrow.zip
# So that they're unique when multiple are downloaded in the next step
Expand Down
2 changes: 1 addition & 1 deletion .github/workflows/ruby.yml
Original file line number Diff line number Diff line change
Expand Up @@ -349,7 +349,7 @@ jobs:
ARROW_WITH_SNAPPY: ON
ARROW_WITH_ZLIB: ON
ARROW_WITH_ZSTD: ON
CMAKE_CXX_STANDARD: "17"
CMAKE_CXX_STANDARD: "20"
CMAKE_GENERATOR: Ninja
CMAKE_INSTALL_PREFIX: "${{ github.workspace }}/dist"
VCPKG_BINARY_SOURCES: 'clear;nuget,GitHub,readwrite'
Expand Down
2 changes: 1 addition & 1 deletion c_glib/meson.build
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@ project(
'arrow-glib',
'c',
'cpp',
default_options: ['c_std=c99', 'cpp_std=c++17'],
default_options: ['c_std=c99', 'cpp_std=c++20'],
license: 'Apache-2.0',
# Debian:
# https://packages.debian.org/search?keywords=meson
Expand Down
2 changes: 1 addition & 1 deletion ci/appveyor-cpp-build.bat
Original file line number Diff line number Diff line change
Expand Up @@ -93,7 +93,7 @@ cmake -G "%GENERATOR%" %ARROW_CMAKE_ARGS% ^
-DARROW_WITH_ZSTD=ON ^
-DCMAKE_BUILD_TYPE="Release" ^
-DCMAKE_CXX_FLAGS_RELEASE="/MD /Od /UNDEBUG" ^
-DCMAKE_CXX_STANDARD=17 ^
-DCMAKE_CXX_STANDARD=20 ^
-DCMAKE_INSTALL_PREFIX=%CONDA_PREFIX%\Library ^
-DCMAKE_UNITY_BUILD=ON ^
-DCMAKE_VERBOSE_MAKEFILE=OFF ^
Expand Down
5 changes: 3 additions & 2 deletions ci/docker/centos-7-cpp.dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,7 @@ RUN sed -i \
-e 's/mirror\.centos\.org/vault.centos.org/' \
/etc/yum.repos.d/*.repo

# devtoolset is required for C++17
# devtoolset is required for C++20
RUN \
yum install -y \
centos-release-scl \
Expand All @@ -38,7 +38,8 @@ RUN \
/etc/yum.repos.d/CentOS-SCLo-scl*.repo && \
yum install -y \
curl \
devtoolset-8 \
devtoolset-11-gcc-c++ \
devtoolset-11-make \
diffutils \
gcc-c++ \
libcurl-devel \
Expand Down
4 changes: 2 additions & 2 deletions ci/scripts/cpp_build.sh
Original file line number Diff line number Diff line change
Expand Up @@ -167,7 +167,7 @@ elif [ "${ARROW_EMSCRIPTEN:-OFF}" = "ON" ]; then
-DCMAKE_VERBOSE_MAKEFILE=${CMAKE_VERBOSE_MAKEFILE:-OFF} \
-DCMAKE_C_FLAGS="${CFLAGS:-}" \
-DCMAKE_CXX_FLAGS="${CXXFLAGS:-}" \
-DCMAKE_CXX_STANDARD="${CMAKE_CXX_STANDARD:-17}" \
-DCMAKE_CXX_STANDARD="${CMAKE_CXX_STANDARD:-20}" \
-DCMAKE_INSTALL_LIBDIR=${CMAKE_INSTALL_LIBDIR:-lib} \
-DCMAKE_INSTALL_PREFIX=${CMAKE_INSTALL_PREFIX:-${ARROW_HOME}} \
-DCMAKE_UNITY_BUILD=${CMAKE_UNITY_BUILD:-OFF} \
Expand Down Expand Up @@ -255,7 +255,7 @@ else
-DCMAKE_VERBOSE_MAKEFILE=${CMAKE_VERBOSE_MAKEFILE:-OFF} \
-DCMAKE_C_FLAGS="${CFLAGS:-}" \
-DCMAKE_CXX_FLAGS="${CXXFLAGS:-}" \
-DCMAKE_CXX_STANDARD="${CMAKE_CXX_STANDARD:-17}" \
-DCMAKE_CXX_STANDARD="${CMAKE_CXX_STANDARD:-20}" \
-DCMAKE_INSTALL_LIBDIR=${CMAKE_INSTALL_LIBDIR:-lib} \
-DCMAKE_INSTALL_PREFIX=${CMAKE_INSTALL_PREFIX:-${ARROW_HOME}} \
-DCMAKE_UNITY_BUILD=${CMAKE_UNITY_BUILD:-OFF} \
Expand Down
2 changes: 1 addition & 1 deletion ci/scripts/r_windows_build.sh
Original file line number Diff line number Diff line change
Expand Up @@ -44,7 +44,7 @@ mv mingw* build
cd build

# This may vary by system/CI provider
MSYS_LIB_DIR="/c/rtools40"
MSYS_LIB_DIR="/c/rtools${RTOOLS_VERSION}"

# Untar the builds we made
ls *.xz | xargs -n 1 tar -xJf
Expand Down
6 changes: 3 additions & 3 deletions cpp/CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -625,10 +625,10 @@ set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${ARROW_CXXFLAGS}")
# C++ specific flags.
set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${CXX_COMMON_FLAGS} ${ARROW_CXXFLAGS}")

# Remove --std=c++17 to avoid errors from C compilers
string(REPLACE "-std=c++17" "" CMAKE_C_FLAGS ${CMAKE_C_FLAGS})
# Remove -std=c++20 to avoid errors from C compilers
string(REPLACE "-std=c++20" "" CMAKE_C_FLAGS ${CMAKE_C_FLAGS})

# Add C++-only flags, like -std=c++17
# Add C++-only flags, like -std=c++20
set(CMAKE_CXX_FLAGS "${CXX_ONLY_FLAGS} ${CMAKE_CXX_FLAGS}")

# ASAN / TSAN / UBSAN
Expand Down
2 changes: 1 addition & 1 deletion cpp/cmake_modules/GandivaAddBitcode.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@

# Create bitcode for the given source file.
function(gandiva_add_bitcode SOURCE)
set(CLANG_OPTIONS -std=c++17)
set(CLANG_OPTIONS -std=c++20)
if(MSVC)
# "19.20" means that it's compatible with Visual Studio 16 2019.
# We can update this to "19.30" when we dropped support for Visual
Expand Down
8 changes: 4 additions & 4 deletions cpp/cmake_modules/SetupCxxFlags.cmake
Original file line number Diff line number Diff line change
Expand Up @@ -137,12 +137,12 @@ endif()

# This ensures that things like c++17 get passed correctly
if(NOT DEFINED CMAKE_CXX_STANDARD)
set(CMAKE_CXX_STANDARD 17)
elseif(${CMAKE_CXX_STANDARD} VERSION_LESS 17)
message(FATAL_ERROR "Cannot set a CMAKE_CXX_STANDARD smaller than 17")
set(CMAKE_CXX_STANDARD 20)
elseif(${CMAKE_CXX_STANDARD} VERSION_LESS 20)
message(FATAL_ERROR "Cannot set a CMAKE_CXX_STANDARD smaller than 20")
endif()

# We require a C++17 compliant compiler
# We require a C++20 compliant compiler
set(CMAKE_CXX_STANDARD_REQUIRED ON)

# ARROW-6848: Do not use GNU (or other CXX) extensions
Expand Down
4 changes: 2 additions & 2 deletions cpp/examples/minimal_build/CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -24,10 +24,10 @@ option(ARROW_LINK_SHARED "Link to the Arrow shared library" ON)
find_package(Arrow REQUIRED)

if(NOT DEFINED CMAKE_CXX_STANDARD)
set(CMAKE_CXX_STANDARD 17)
set(CMAKE_CXX_STANDARD 20)
endif()

# We require a C++17 compliant compiler
# We require a C++20 compliant compiler
set(CMAKE_CXX_STANDARD_REQUIRED ON)

if(NOT CMAKE_BUILD_TYPE)
Expand Down
2 changes: 1 addition & 1 deletion cpp/examples/minimal_build/run_static.sh
Original file line number Diff line number Diff line change
Expand Up @@ -86,7 +86,7 @@ echo

rm -rf $EXAMPLE_BUILD_DIR
mkdir -p $EXAMPLE_BUILD_DIR
${CXX:-c++} -std=c++17 \
${CXX:-c++} -std=c++20 \
-o $EXAMPLE_BUILD_DIR/arrow-example \
$EXAMPLE_DIR/example.cc \
$(PKG_CONFIG_PATH=$ARROW_BUILD_DIR/lib/pkgconfig \
Expand Down
4 changes: 2 additions & 2 deletions cpp/examples/parquet/parquet_arrow/CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -28,10 +28,10 @@ option(PARQUET_LINK_SHARED "Link to the Parquet shared library" ON)

# This ensures that things like -std=gnu++... get passed correctly
if(NOT DEFINED CMAKE_CXX_STANDARD)
set(CMAKE_CXX_STANDARD 17)
set(CMAKE_CXX_STANDARD 20)
endif()

# We require a C++17 compliant compiler
# We require a C++20 compliant compiler
set(CMAKE_CXX_STANDARD_REQUIRED ON)

# Look for installed packages the system
Expand Down
2 changes: 1 addition & 1 deletion cpp/examples/tutorial_examples/CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,7 @@ project(ArrowTutorialExamples)

find_package(ArrowDataset)

set(CMAKE_CXX_STANDARD 17)
set(CMAKE_CXX_STANDARD 20)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Werror -Wall -Wextra")

set(CMAKE_BUILD_TYPE Release)
Expand Down
2 changes: 1 addition & 1 deletion cpp/meson.build
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,7 @@ project(
'buildtype=release',
'c_std=c99',
'warning_level=2',
'cpp_std=c++17',
'cpp_std=c++20',
],
)

Expand Down
102 changes: 66 additions & 36 deletions cpp/src/arrow/array/array_nested.cc
Original file line number Diff line number Diff line change
Expand Up @@ -17,10 +17,10 @@

#include "arrow/array/array_nested.h"

#include <atomic>
#include <cstddef>
#include <cstdint>
#include <memory>
#include <mutex>
#include <string>
#include <utility>
#include <vector>
Expand Down Expand Up @@ -1004,10 +1004,18 @@ Result<std::shared_ptr<Array>> FixedSizeListArray::Flatten(
// ----------------------------------------------------------------------
// Struct

struct StructArray::Impl {
std::mutex mutex;
ArrayVector boxed_fields;
};

StructArray::~StructArray() = default;

StructArray::StructArray(const std::shared_ptr<ArrayData>& data) {
ARROW_CHECK_EQ(data->type->id(), Type::STRUCT);
SetData(data);
boxed_fields_.resize(data->child_data.size());
impl_ = std::make_unique<Impl>();
impl_->boxed_fields.resize(data_->child_data.size());
}

StructArray::StructArray(const std::shared_ptr<DataType>& type, int64_t length,
Expand All @@ -1016,10 +1024,12 @@ StructArray::StructArray(const std::shared_ptr<DataType>& type, int64_t length,
int64_t offset) {
ARROW_CHECK_EQ(type->id(), Type::STRUCT);
SetData(ArrayData::Make(type, length, {std::move(null_bitmap)}, null_count, offset));
data_->child_data.reserve(children.size());
for (const auto& child : children) {
data_->child_data.push_back(child->data());
}
boxed_fields_.resize(children.size());
impl_ = std::make_unique<Impl>();
impl_->boxed_fields.resize(data_->child_data.size());
}

Result<std::shared_ptr<StructArray>> StructArray::Make(
Expand Down Expand Up @@ -1069,27 +1079,32 @@ const StructType* StructArray::struct_type() const {
return checked_cast<const StructType*>(data_->type.get());
}

std::shared_ptr<Array> StructArray::MakeBoxedField(int i) const {
std::shared_ptr<ArrayData> field_data;
if (data_->offset != 0 || data_->child_data[i]->length != data_->length) {
field_data = data_->child_data[i]->Slice(data_->offset, data_->length);
} else {
field_data = data_->child_data[i];
}
return MakeArray(field_data);
}

const ArrayVector& StructArray::fields() const {
std::lock_guard lock(impl_->mutex);
for (int i = 0; i < num_fields(); ++i) {
(void)field(i);
if (impl_->boxed_fields[i] == nullptr) {
impl_->boxed_fields[i] = MakeBoxedField(i);
}
}
return boxed_fields_;
return impl_->boxed_fields;
}

const std::shared_ptr<Array>& StructArray::field(int i) const {
std::shared_ptr<Array> result = std::atomic_load(&boxed_fields_[i]);
if (!result) {
std::shared_ptr<ArrayData> field_data;
if (data_->offset != 0 || data_->child_data[i]->length != data_->length) {
field_data = data_->child_data[i]->Slice(data_->offset, data_->length);
} else {
field_data = data_->child_data[i];
}
result = MakeArray(field_data);
std::atomic_store(&boxed_fields_[i], std::move(result));
return boxed_fields_[i];
std::lock_guard lock(impl_->mutex);
if (impl_->boxed_fields[i] == nullptr) {
impl_->boxed_fields[i] = MakeBoxedField(i);
}
return boxed_fields_[i];
return impl_->boxed_fields[i];
}

std::shared_ptr<Array> StructArray::GetFieldByName(const std::string& name) const {
Expand Down Expand Up @@ -1177,14 +1192,24 @@ Result<std::shared_ptr<Array>> StructArray::GetFlattenedField(int index,
// ----------------------------------------------------------------------
// UnionArray

struct UnionArray::Impl {
std::mutex mutex;
ArrayVector boxed_fields;
};

UnionArray::UnionArray() = default;
UnionArray::~UnionArray() = default;

void UnionArray::SetData(std::shared_ptr<ArrayData> data) {
this->Array::SetData(std::move(data));

union_type_ = checked_cast<const UnionType*>(data_->type.get());

ARROW_CHECK_GE(data_->buffers.size(), 2);
raw_type_codes_ = data->GetValuesSafe<int8_t>(1);
boxed_fields_.resize(data_->child_data.size());

impl_ = std::make_unique<Impl>();
impl_->boxed_fields.resize(data_->child_data.size());
}

void SparseUnionArray::SetData(std::shared_ptr<ArrayData> data) {
Expand All @@ -1198,16 +1223,16 @@ void SparseUnionArray::SetData(std::shared_ptr<ArrayData> data) {

void DenseUnionArray::SetData(const std::shared_ptr<ArrayData>& data) {
this->UnionArray::SetData(data);

ARROW_CHECK_EQ(data_->type->id(), Type::DENSE_UNION);
ARROW_CHECK_EQ(data_->buffers.size(), 3);

// No validity bitmap
ARROW_CHECK_EQ(data_->buffers[0], nullptr);

raw_value_offsets_ = data->GetValuesSafe<int32_t>(2);
}

SparseUnionArray::~SparseUnionArray() = default;

SparseUnionArray::SparseUnionArray(std::shared_ptr<ArrayData> data) {
SetData(std::move(data));
}
Expand Down Expand Up @@ -1261,6 +1286,8 @@ Result<std::shared_ptr<Array>> SparseUnionArray::GetFlattenedField(
return MakeArray(child_data);
}

DenseUnionArray::~DenseUnionArray() = default;

DenseUnionArray::DenseUnionArray(const std::shared_ptr<ArrayData>& data) {
SetData(data);
}
Expand Down Expand Up @@ -1352,26 +1379,29 @@ Result<std::shared_ptr<Array>> SparseUnionArray::Make(
return std::make_shared<SparseUnionArray>(std::move(internal_data));
}

std::shared_ptr<Array> UnionArray::MakeBoxedField(int i) const {
std::shared_ptr<ArrayData> child_data = data_->child_data[i]->Copy();
if (mode() == UnionMode::SPARSE) {
// Sparse union: need to adjust child if union is sliced
// (for dense unions, the need to lookup through the offsets
// makes this unnecessary)
if (data_->offset != 0 || child_data->length > data_->length) {
child_data = child_data->Slice(data_->offset, data_->length);
}
}
return MakeArray(child_data);
}

std::shared_ptr<Array> UnionArray::field(int i) const {
if (i < 0 ||
static_cast<decltype(boxed_fields_)::size_type>(i) >= boxed_fields_.size()) {
if (i < 0 || i >= num_fields()) {
return nullptr;
}
std::shared_ptr<Array> result = std::atomic_load(&boxed_fields_[i]);
if (!result) {
std::shared_ptr<ArrayData> child_data = data_->child_data[i]->Copy();
if (mode() == UnionMode::SPARSE) {
// Sparse union: need to adjust child if union is sliced
// (for dense unions, the need to lookup through the offsets
// makes this unnecessary)
if (data_->offset != 0 || child_data->length > data_->length) {
child_data = child_data->Slice(data_->offset, data_->length);
}
}
result = MakeArray(child_data);
std::atomic_store(&boxed_fields_[i], result);

std::lock_guard lock(impl_->mutex);
if (impl_->boxed_fields[i] == nullptr) {
impl_->boxed_fields[i] = MakeBoxedField(i);
}
return result;
return impl_->boxed_fields[i];
}

} // namespace arrow
Loading
Loading