[C++] vcpkg arrow port ArrowDataset component usage

[asfimport]

vcpkg arrow port (https://github.com/microsoft/vcpkg/tree/master/ports/arrow) installs not only arrow but also other arrow components such as ArrowDataset/ArrowFlight. After installation, I can use find_package(Arrow) in CMake/vcpkg to find the arrow package and use it. But I wonder how I can use other components such as ArrowDataset. I tried using it in CMake like below:

1) find_package(ArrowDataset REQUIRED)

CMake will report error like below:

– _find_package pkg_name=ArrowDataset args=REQUIRED
CMake Error at /Users/ss/dev/tools/vcpkg/scripts/buildsystems/vcpkg.cmake:789 (_find_package):
By not providing "FindArrowDataset.cmake" in CMAKE_MODULE_PATH this project
has asked CMake to find a package configuration file provided by
"ArrowDataset", but CMake did not find one.

Could not find a package configuration file provided by "ArrowDataset" with
any of the following names:

ArrowDatasetConfig.cmake
arrowdataset-config.cmake

Add the installation prefix of "ArrowDataset" to CMAKE_PREFIX_PATH or set
"ArrowDataset_DIR" to a directory containing one of the above files.  If
"ArrowDataset" provides a separate development package or SDK, be sure it
has been installed.
 

2. find_package(Arrow CONFIGS ArrowDatasetConfig.cmake REQUIRED)

The ArrowDatasetConfig.cmake can be found, however, it tries to find Parquet internally and still failed with error like below:

– _find_package pkg_name=Arrow args=CONFIGS;ArrowDatasetConfig.cmake;REQUIRED
– _find_package pkg_name=Arrow args=REQUIRED
– _find_package pkg_name=Threads args=REQUIRED
– _find_package pkg_name=Snappy args=REQUIRED
– _find_package pkg_name=PkgConfig args=QUIET
– _find_package pkg_name=Threads args=REQUIRED
– _find_package pkg_name=Thrift args=REQUIRED
– Found thrift: /Users/ss/dev/tools/vcpkg/installed/x64-osx
– _find_package pkg_name=Threads args=QUIET
– _find_package pkg_name=lz4 args=REQUIRED
– _find_package pkg_name=zstd args=REQUIRED
– _find_package pkg_name=re2Alt args=REQUIRED
– _find_package pkg_name=re2 args=
– _find_package pkg_name=BZip2 args=REQUIRED
– _find_package pkg_name=unofficial-utf8proc args=REQUIRED
– _find_package pkg_name=c-ares args=
– _find_package pkg_name=c-ares args=CONFIG
– _find_package pkg_name=absl args=CONFIG
– _find_package pkg_name=Threads args=
– _find_package pkg_name=upb args=CONFIG;REQUIRED
– _find_package pkg_name=Parquet args=REQUIRED
CMake Error at /Users/ss/dev/tools/vcpkg/scripts/buildsystems/vcpkg.cmake:789 (_find_package):
By not providing "FindParquet.cmake" in CMAKE_MODULE_PATH this project has
asked CMake to find a package configuration file provided by "Parquet", but
CMake did not find one.

Could not find a package configuration file provided by "Parquet" with any
of the following names:

ParquetConfig.cmake
parquet-config.cmake

Add the installation prefix of "Parquet" to CMAKE_PREFIX_PATH or set
"Parquet_DIR" to a directory containing one of the above files.  If
"Parquet" provides a separate development package or SDK, be sure it has
been installed.
Call Stack (most recent call first):
/usr/local/Cellar/cmake/3.21.3_1/share/cmake/Modules/CMakeFindDependencyMacro.cmake:47 (find_package)
/Users/ss/dev/tools/vcpkg/installed/x64-osx/share/arrow/ArrowDatasetConfig.cmake:55 (find_dependency)
/Users/ss/dev/tools/vcpkg/scripts/buildsystems/vcpkg.cmake:789 (_find_package)
Is there any instruction describing how this could be done? Thanks.

Reporter: Yue Ni / @niyue

Related issues:

• [C++] CMake's find_package(Parquet) does not find Parquet with Arrow 3.0.0 (is duplicated by)

_{Note: This issue was originally created as ARROW-14681. Please see the migration documentation for further details.}

[asfimport]

Kouhei Sutou / @kou:
This is an our CMake package problem.
We hope that we can fix this in the next release.

Workaround:

find_package(Arrow REQUIRED)
get_filename_component(ARROW_SEARCH_DIR ${Arrow_CONFIG} DIRECTORY)
find_package(ArrowDataset REQUIRED HINTS ${ARROW_SEARCH_DIR})

See also: ARROW-12175

[asfimport]

Kouhei Sutou / @kou:
We can use find_package(ArrowDataset) with Apache Arrow 10.0.0.
If there is a problem with it, please open a new issue at https://github.com/apache/arrow/issues/new/choose .