GH-46231: [C++][CMake] Fix arrow_bundled_dependencies to be externally accessible by FetchContent

[wgtmac]

Rationale for this change

arrow_bundled_dependencies is an IMPORTED library which links to arrow_bundled_dependencies_merge. When arrow-cpp is used externally as a third-party dependency via CMake FetchContent, target arrow_bundled_dependencies is not accessible by the parent project (but arrow_bundled_dependencies_merge is). It is difficult for the parent project to locate the built product of arrow_bundled_dependencies_merge and install it.

What changes are included in this PR?

Change IMPORTED library arrow_bundled_dependencies to be GLOBAL.

Are these changes tested?

• Pass all CIs.
• Verified to work by feat: enable arrow to build parquet iceberg-cpp#89

Are there any user-facing changes?

I don't think so.

• GitHub Issue: [C++][CMake] arrow_bundled_dependencies is non-accessible when used externally via FetchContent #46231

[github-actions]

⚠️ GitHub issue #46231 has been automatically assigned in GitHub to PR creator.

[wgtmac]

@kou @raulcd Could you help review this?

cc @lidavidm

[kou]

@github-actions crossbow submit -g cpp -g r

[kou]

+1 if CI is green.

FYI: We don't support FetchContent yet...:

• #43668
• #32234

[github-actions]

Revision: 7d50c4a

Submitted crossbow builds: ursacomputing/crossbow @ actions-7cc39ab5b3

Task	Status
example-cpp-minimal-build-static
example-cpp-minimal-build-static-system-dependency
example-cpp-tutorial
r-binary-packages
r-recheck-most
test-alpine-linux-cpp
test-build-cpp-fuzz
test-conda-cpp
test-conda-cpp-meson
test-conda-cpp-valgrind
test-cuda-cpp-ubuntu-22.04-cuda-11.7.1
test-debian-12-cpp-amd64
test-debian-12-cpp-i386
test-fedora-39-cpp
test-r-arrow-backwards-compatibility
test-r-clang-asan
test-r-clang-ubsan
test-r-depsource-bundled
test-r-depsource-system
test-r-dev-duckdb
test-r-devdocs
test-r-extra-packages
test-r-gcc-11
test-r-gcc-12
test-r-install-local
test-r-install-local-minsizerel
test-r-linux-as-cran
test-r-linux-rchk
test-r-linux-sanitizer
test-r-linux-valgrind
test-r-m1-san
test-r-macos-as-cran
test-r-minimal-build
test-r-offline-maximal
test-r-offline-minimal
test-r-rhub-debian-gcc-devel-lto-latest
test-r-rhub-debian-gcc-release-custom-ccache
test-r-rhub-ubuntu-release-latest
test-r-rocker-r-ver-latest
test-r-rstudio-r-base-4.1-opensuse155
test-r-rstudio-r-base-4.2-focal
test-r-ubuntu-22.04
test-r-versions
test-ubuntu-22.04-cpp
test-ubuntu-22.04-cpp-20
test-ubuntu-22.04-cpp-bundled
test-ubuntu-22.04-cpp-emscripten
test-ubuntu-22.04-cpp-no-threading
test-ubuntu-24.04-cpp
test-ubuntu-24.04-cpp-bundled-offline
test-ubuntu-24.04-cpp-gcc-13-bundled
test-ubuntu-24.04-cpp-gcc-14
test-ubuntu-24.04-cpp-minimal-with-formats
test-ubuntu-24.04-cpp-thread-sanitizer

[wgtmac]

FYI: We don't support FetchContent yet...:

It seems that I managed to use FetchContent to import Arrow in the iceberg-cpp. Did I miss something?

[wgtmac]

Should we merge this? @kou

[kou]

It seems that I managed to use FetchContent to import Arrow in the iceberg-cpp. Did I miss something?

We may need more changes for full FetchContent support. We don't have enough check for it. For example, we should use FetchContent instead of ExternalProject_Add() for MATLAB:

arrow/matlab/CMakeLists.txt

Lines 20 to 99 in 8e61cdd

	# Build the Arrow C++ libraries using ExternalProject_Add.
	function(build_arrow)
	set(options)
	set(one_value_args)
	set(multi_value_args)
	cmake_parse_arguments(ARG
	"${options}"
	"${one_value_args}"
	"${multi_value_args}"
	${ARGN})
	if(ARG_UNPARSED_ARGUMENTS)
	message(SEND_ERROR "Error: unrecognized arguments: ${ARG_UNPARSED_ARGUMENTS}")
	endif()
	set(ARROW_PREFIX "${CMAKE_CURRENT_BINARY_DIR}/arrow_ep-prefix")
	set(ARROW_BINARY_DIR "${CMAKE_CURRENT_BINARY_DIR}/arrow_ep-build")
	# Supply -DARROW_CXXFLAGS=-D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR to fix
	# a segfault on windows. See https://github.com/apache/arrow/issues/42015.
	set(ARROW_CMAKE_ARGS
	"-DCMAKE_INSTALL_PREFIX=${ARROW_PREFIX}"
	"-DCMAKE_INSTALL_LIBDIR=lib"
	"-DARROW_BUILD_STATIC=OFF"
	"-DARROW_CSV=ON"
	"-DARROW_CXXFLAGS=-D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR")
	add_library(arrow_shared SHARED IMPORTED)
	set(ARROW_LIBRARY_TARGET arrow_shared)
	# Set the runtime shared library (.dll, .so, or .dylib)
	if(WIN32)
	# The shared library (i.e. .dll) is located in the "bin" directory.
	set(ARROW_SHARED_LIBRARY_DIR "${ARROW_PREFIX}/bin")
	else()
	# The shared library (i.e. .so or .dylib) is located in the "lib" directory.
	set(ARROW_SHARED_LIBRARY_DIR "${ARROW_PREFIX}/lib")
	endif()
	set(ARROW_SHARED_LIB_FILENAME
	"${CMAKE_SHARED_LIBRARY_PREFIX}arrow${CMAKE_SHARED_LIBRARY_SUFFIX}")
	set(ARROW_SHARED_LIB "${ARROW_SHARED_LIBRARY_DIR}/${ARROW_SHARED_LIB_FILENAME}")
	set_target_properties(arrow_shared PROPERTIES IMPORTED_LOCATION ${ARROW_SHARED_LIB})
	# Set the link-time import library (.lib)
	if(WIN32)
	# The import library (i.e. .lib) is located in the "lib" directory.
	set(ARROW_IMPORT_LIB_FILENAME
	"${CMAKE_IMPORT_LIBRARY_PREFIX}arrow${CMAKE_IMPORT_LIBRARY_SUFFIX}")
	set(ARROW_IMPORT_LIB "${ARROW_PREFIX}/lib/${ARROW_IMPORT_LIB_FILENAME}")
	set_target_properties(arrow_shared PROPERTIES IMPORTED_IMPLIB ${ARROW_IMPORT_LIB})
	endif()
	# Set the include directories
	set(ARROW_INCLUDE_DIR "${ARROW_PREFIX}/include")
	file(MAKE_DIRECTORY "${ARROW_INCLUDE_DIR}")
	set_target_properties(arrow_shared PROPERTIES INTERFACE_INCLUDE_DIRECTORIES
	${ARROW_INCLUDE_DIR})
	# Set the build byproducts for the ExternalProject build
	if(WIN32)
	set(ARROW_BUILD_BYPRODUCTS "${ARROW_IMPORT_LIB}")
	else()
	set(ARROW_BUILD_BYPRODUCTS "${ARROW_SHARED_LIB}")
	endif()
	# Building the Arrow C++ libraries requires ExternalProject.
	include(ExternalProject)
	externalproject_add(arrow_ep
	SOURCE_DIR "${CMAKE_SOURCE_DIR}/../cpp"
	BINARY_DIR "${ARROW_BINARY_DIR}"
	CMAKE_ARGS "${ARROW_CMAKE_ARGS}"
	BUILD_BYPRODUCTS "${ARROW_BUILD_BYPRODUCTS}")
	add_dependencies(${ARROW_LIBRARY_TARGET} arrow_ep)
	endfunction()

[wgtmac]

Yes, I only used limited features from Arrow and Parquet. If I need to use provided libraries like zstd and zlib, I may hit new issues.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 8e61cdd.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 1 possible false positive for unstable benchmarks that are known to sometimes produce them.