Merge branch 'main' into add-checksum-rbinaries

.env

@@ -54,7 +54,7 @@ UBUNTU=20.04
 
 # Default versions for various dependencies
 CLANG_TOOLS=14
-CUDA=11.0.3
+CUDA=11.2.2
 DASK=latest
 DOTNET=7.0
 GCC_VERSION=""

LICENSE.txt

@@ -1889,19 +1889,6 @@ for PyArrow. Ibis is released under the Apache License, Version 2.0.
 
 --------------------------------------------------------------------------------
 
-This project includes code from the autobrew project.
-
-The following files are based on code from the autobrew project:
-* r/tools/autobrew
-* dev/tasks/homebrew-formulae/autobrew/apache-arrow.rb
-* dev/tasks/homebrew-formulae/autobrew/apache-arrow-static.rb 
-
-Copyright (c) 2019, Jeroen Ooms
-License: MIT
-Homepage: https://github.com/autobrew/
-
---------------------------------------------------------------------------------
-
 dev/tasks/homebrew-formulae/apache-arrow.rb has the following license:
 
 BSD 2-Clause License

c_glib/test/dataset/test-file-system-dataset-factory.rb

@@ -109,9 +109,9 @@ def test_validate_fragments
                                     point: Arrow::Int16DataType.new)
       options.validate_fragments = true
       message = "[file-system-dataset-factory][finish]: " +
-                "Invalid: Unable to merge: " +
+                "Type error: Unable to merge: " +
                 "Field point has incompatible types: int16 vs int32"
-      error = assert_raise(Arrow::Error::Invalid) do
+      error = assert_raise(Arrow::Error::Type) do
         @factory.finish(options)
       end
       assert_equal(message, error.message.lines(chomp: true).first)

ci/docker/ubuntu-22.04-cpp.dockerfile

@@ -79,6 +79,7 @@ RUN apt-get update -y -q && \
         libc-ares-dev \
         libcurl4-openssl-dev \
         libgflags-dev \
+        libgmock-dev \
         libgoogle-glog-dev \
         libgrpc++-dev \
         libidn2-dev \
@@ -166,7 +167,6 @@ RUN /arrow/ci/scripts/install_sccache.sh unknown-linux-musl /usr/local/bin
 # provided by the distribution:
 # - Abseil is old
 # - libc-ares-dev does not install CMake config files
-# - libgtest-dev only provide sources
 ENV absl_SOURCE=BUNDLED \
     ARROW_ACERO=ON \
     ARROW_AZURE=ON \
@@ -200,7 +200,6 @@ ENV absl_SOURCE=BUNDLED \
     AWSSDK_SOURCE=BUNDLED \
     Azure_SOURCE=BUNDLED \
     google_cloud_cpp_storage_SOURCE=BUNDLED \
-    GTest_SOURCE=BUNDLED \
     ORC_SOURCE=BUNDLED \
     PARQUET_BUILD_EXAMPLES=ON \
     PARQUET_BUILD_EXECUTABLES=ON \

ci/scripts/cpp_build.sh

@@ -53,6 +53,16 @@ if [ "${GITHUB_ACTIONS:-false}" = "true" ]; then
   esac
 fi
 
+if [ "${ARROW_ENABLE_THREADING:-ON}" = "OFF" ]; then
+  ARROW_FLIGHT=OFF
+  ARROW_FLIGHT_SQL=OFF
+  ARROW_GCS=OFF
+  ARROW_JEMALLOC=OFF
+  ARROW_MIMALLOC=OFF
+  ARROW_S3=OFF
+  ARROW_WITH_OPENTELEMETRY=OFF
+fi
+
 if [ "${ARROW_USE_CCACHE}" == "ON" ]; then
     echo -e "===\n=== ccache statistics before build\n==="
     ccache -sv 2>/dev/null || ccache -s

ci/scripts/go_build.sh

@@ -50,13 +50,13 @@ if [[ -n "${ARROW_GO_INTEGRATION}" ]]; then
             go_lib="arrow_go_integration.so"
             ;;
         Darwin)
-            go_lib="arrow_go_integration.so"
+            go_lib="arrow_go_integration.dylib"
             ;;
         MINGW*)
             go_lib="arrow_go_integration.dll"
             ;;
     esac
-    go build -tags cdata_integration,assert -buildmode=c-shared -o ${go_lib} .
+    go build -buildvcs=false -tags cdata_integration,assert -buildmode=c-shared -o ${go_lib} .
 
     popd
 fi

ci/scripts/integration_dask.sh

@@ -34,10 +34,7 @@ python -c "import dask.dataframe"
 # The "skip_with_pyarrow_strings" marker is meant to skip automatically, but that doesn't work with --pyargs, so de-selecting manually
 pytest -v --pyargs dask.dataframe.tests.test_dataframe -m "not skip_with_pyarrow_strings"
 pytest -v --pyargs dask.dataframe.io.tests.test_orc
-# skip failing parquet tests
-# test_pandas_timestamp_overflow_pyarrow is skipped because of GH-33321.
 pytest -v --pyargs dask.dataframe.io.tests.test_parquet \
-  -k "not test_pandas_timestamp_overflow_pyarrow" \
   -m "not skip_with_pyarrow_strings and not xfail_with_pyarrow_strings"
 # this file contains parquet tests that use S3 filesystem
 pytest -v --pyargs dask.bytes.tests.test_s3

cpp/CMakeLists.txt

@@ -961,7 +961,9 @@ endif()
 
 set(ARROW_SYSTEM_LINK_LIBS)
 
-list(APPEND ARROW_SYSTEM_LINK_LIBS Threads::Threads)
+if(ARROW_ENABLE_THREADING)
+  list(APPEND ARROW_SYSTEM_LINK_LIBS Threads::Threads)
+endif()
 if(CMAKE_THREAD_LIBS_INIT)
   string(APPEND ARROW_PC_LIBS_PRIVATE " ${CMAKE_THREAD_LIBS_INIT}")
 endif()

cpp/cmake_modules/DefineOptions.cmake

@@ -340,12 +340,16 @@ takes precedence over ccache if a storage backend is configured" ON)
   define_option(ARROW_IPC "Build the Arrow IPC extensions" ON)
 
   set(ARROW_JEMALLOC_DESCRIPTION "Build the Arrow jemalloc-based allocator")
-  if(WIN32 OR "${CMAKE_SYSTEM_NAME}" STREQUAL "FreeBSD")
+  if(WIN32
+     OR "${CMAKE_SYSTEM_NAME}" STREQUAL "FreeBSD"
+     OR NOT ARROW_ENABLE_THREADING)
     # jemalloc is not supported on Windows.
     #
     # jemalloc is the default malloc implementation on FreeBSD and can't
     # be built with --disable-libdl on FreeBSD. Because lazy-lock feature
     # is required on FreeBSD. Lazy-lock feature requires libdl.
+    #
+    # jemalloc requires thread.
     define_option(ARROW_JEMALLOC ${ARROW_JEMALLOC_DESCRIPTION} OFF)
   else()
     define_option(ARROW_JEMALLOC ${ARROW_JEMALLOC_DESCRIPTION} ON)