Changes to compile with Clang15

CMake/ResolveDependency.cmake

@@ -40,6 +40,8 @@ set(CMAKE_TLS_VERIFY true)
 
 macro(build_dependency dependency_name)
   string(TOLOWER ${dependency_name} dependency_name_lower)
+  set(CMAKE_COMPILER_SELECTION_ARGS -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
+                                    -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER})
   include(${dependency_name_lower})
 endmacro()
 

CMake/resolve_dependency_modules/absl.cmake

@@ -28,7 +28,8 @@ FetchContent_Declare(
   absl
   URL ${VELOX_ABSL_SOURCE_URL}
   URL_HASH ${VELOX_ABSL_BUILD_SHA256_CHECKSUM}
-  OVERRIDE_FIND_PACKAGE EXCLUDE_FROM_ALL SYSTEM)
+  OVERRIDE_FIND_PACKAGE EXCLUDE_FROM_ALL SYSTEM CMAKE_ARGS
+  ${CMAKE_COMPILER_SELECTION_ARGS})
 
 set(ABSL_BUILD_TESTING OFF)
 set(ABSL_PROPAGATE_CXX_STD ON)

CMake/resolve_dependency_modules/arrow/CMakeLists.txt

@@ -38,7 +38,8 @@ if(VELOX_ENABLE_ARROW)
       -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}
       -DARROW_BUILD_STATIC=ON
       -DThrift_SOURCE=${THRIFT_SOURCE}
-      -DCMAKE_PREFIX_PATH=${CMAKE_PREFIX_PATH})
+      -DCMAKE_PREFIX_PATH=${CMAKE_PREFIX_PATH}
+      ${CMAKE_COMPILER_SELECTION_ARGS})
   set(ARROW_LIBDIR ${ARROW_PREFIX}/install/${CMAKE_INSTALL_LIBDIR})
 
   add_library(thrift STATIC IMPORTED GLOBAL)

CMake/resolve_dependency_modules/boost/CMakeLists.txt

@@ -42,7 +42,8 @@ endif()
 FetchContent_Declare(
   Boost
   URL ${VELOX_BOOST_SOURCE_URL}
-  URL_HASH ${VELOX_BOOST_BUILD_SHA256_CHECKSUM})
+  URL_HASH ${VELOX_BOOST_BUILD_SHA256_CHECKSUM}
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 # Configure the file before adding the header only libs
 configure_file(${CMAKE_CURRENT_LIST_DIR}/FindBoost.cmake.in

CMake/resolve_dependency_modules/cpr.cmake

@@ -26,15 +26,15 @@ set(curl_SOURCE BUNDLED)
 resolve_dependency(curl)
 
 resolve_dependency_url(CPR)
-
 message(STATUS "Building cpr from source")
 FetchContent_Declare(
   cpr
   URL ${VELOX_CPR_SOURCE_URL}
   URL_HASH ${VELOX_CPR_BUILD_SHA256_CHECKSUM}
   PATCH_COMMAND
     git apply ${CMAKE_CURRENT_LIST_DIR}/cpr/cpr-libcurl-compatible.patch && git
-    apply ${CMAKE_CURRENT_LIST_DIR}/cpr/cpr-remove-sancheck.patch)
+    apply ${CMAKE_CURRENT_LIST_DIR}/cpr/cpr-remove-sancheck.patch CMAKE_ARGS
+    ${CMAKE_COMPILER_SELECTION_ARGS})
 set(BUILD_SHARED_LIBS OFF)
 set(CPR_USE_SYSTEM_CURL OFF)
 # ZLIB has already been found by find_package(ZLIB, REQUIRED), set CURL_ZLIB=OFF

CMake/resolve_dependency_modules/duckdb.cmake

@@ -34,7 +34,8 @@ FetchContent_Declare(
   PATCH_COMMAND
     git apply ${CMAKE_CURRENT_LIST_DIR}/duckdb/remove-ccache.patch && git apply
     ${CMAKE_CURRENT_LIST_DIR}/duckdb/fix-duckdbversion.patch && git apply
-    ${CMAKE_CURRENT_LIST_DIR}/duckdb/re2.patch)
+    ${CMAKE_CURRENT_LIST_DIR}/duckdb/re2.patch CMAKE_ARGS
+    ${CMAKE_COMPILER_SELECTION_ARGS})
 
 set(BUILD_UNITTESTS OFF)
 set(ENABLE_SANITIZER OFF)

CMake/resolve_dependency_modules/fmt.cmake

@@ -25,7 +25,8 @@ message(STATUS "Building fmt from source")
 FetchContent_Declare(
   fmt
   URL ${VELOX_FMT_SOURCE_URL}
-  URL_HASH ${VELOX_FMT_BUILD_SHA256_CHECKSUM})
+  URL_HASH ${VELOX_FMT_BUILD_SHA256_CHECKSUM}
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 # Force fmt to create fmt-config.cmake which can be found by other dependecies
 # (e.g. folly)
 set(FMT_INSTALL ON)

CMake/resolve_dependency_modules/folly/CMakeLists.txt

@@ -34,7 +34,7 @@ FetchContent_Declare(
   URL ${VELOX_FOLLY_SOURCE_URL}
   URL_HASH ${VELOX_FOLLY_BUILD_SHA256_CHECKSUM}
   PATCH_COMMAND git apply ${CMAKE_CURRENT_LIST_DIR}/folly-no-export.patch
-                ${glog_patch})
+                ${glog_patch} CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 if(ON_APPLE_M1)
   # folly will wrongly assume x86_64 if this is not set

CMake/resolve_dependency_modules/gflags.cmake

@@ -27,7 +27,8 @@ FetchContent_Declare(
   gflags
   URL ${VELOX_GFLAGS_SOURCE_URL}
   URL_HASH ${VELOX_GFLAGS_BUILD_SHA256_CHECKSUM}
-  PATCH_COMMAND git apply ${CMAKE_CURRENT_LIST_DIR}/gflags/gflags-config.patch)
+  PATCH_COMMAND git apply ${CMAKE_CURRENT_LIST_DIR}/gflags/gflags-config.patch
+                CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 set(GFLAGS_BUILD_STATIC_LIBS ON)
 set(GFLAGS_BUILD_gflags_LIB ON)

CMake/resolve_dependency_modules/glog.cmake

@@ -27,8 +27,10 @@ FetchContent_Declare(
   glog
   URL ${VELOX_GLOG_SOURCE_URL}
   URL_HASH ${VELOX_GLOG_BUILD_SHA256_CHECKSUM}
-  PATCH_COMMAND git apply ${CMAKE_CURRENT_LIST_DIR}/glog/glog-no-export.patch
-                && git apply ${CMAKE_CURRENT_LIST_DIR}/glog/glog-config.patch)
+  PATCH_COMMAND
+    git apply ${CMAKE_CURRENT_LIST_DIR}/glog/glog-no-export.patch && git apply
+    ${CMAKE_CURRENT_LIST_DIR}/glog/glog-config.patch CMAKE_ARGS
+    ${CMAKE_COMPILER_SELECTION_ARGS})
 
 set(BUILD_SHARED_LIBS OFF)
 set(WITH_UNWIND OFF)

CMake/resolve_dependency_modules/google_cloud_cpp_storage.cmake

@@ -32,7 +32,8 @@ FetchContent_Declare(
   google_cloud_cpp
   URL ${VELOX_GOOGLE_CLOUD_CPP_SOURCE_URL}
   URL_HASH ${VELOX_GOOGLE_CLOUD_CPP_BUILD_SHA256_CHECKSUM}
-  OVERRIDE_FIND_PACKAGE EXCLUDE_FROM_ALL SYSTEM)
+  OVERRIDE_FIND_PACKAGE EXCLUDE_FROM_ALL SYSTEM CMAKE_ARGS
+  ${CMAKE_COMPILER_SELECTION_ARGS})
 
 set(GOOGLE_CLOUD_CPP_ENABLE_EXAMPLES OFF)
 set(GOOGLE_CLOUD_CPP_ENABLE

CMake/resolve_dependency_modules/grpc.cmake

@@ -31,7 +31,8 @@ FetchContent_Declare(
   gRPC
   URL ${VELOX_GRPC_SOURCE_URL}
   URL_HASH ${VELOX_GRPC_BUILD_SHA256_CHECKSUM}
-  OVERRIDE_FIND_PACKAGE EXCLUDE_FROM_ALL)
+  OVERRIDE_FIND_PACKAGE EXCLUDE_FROM_ALL CMAKE_ARGS
+  ${CMAKE_COMPILER_SELECTION_ARGS})
 
 set(gRPC_ABSL_PROVIDER
     "package"

CMake/resolve_dependency_modules/gtest.cmake

@@ -26,7 +26,8 @@ message(STATUS "Building gtest from source")
 FetchContent_Declare(
   gtest
   URL ${VELOX_GTEST_SOURCE_URL}
-  URL_HASH ${VELOX_GTEST_BUILD_SHA256_CHECKSUM})
+  URL_HASH ${VELOX_GTEST_BUILD_SHA256_CHECKSUM}
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 FetchContent_MakeAvailable(gtest)
 

CMake/resolve_dependency_modules/icu.cmake

@@ -54,7 +54,8 @@ ExternalProject_Add(
   CONFIGURE_COMMAND <SOURCE_DIR>/source/configure --prefix=${ICU_DIR}
                     --libdir=${ICU_LIBRARIES} ${ICU_CFG}
   BUILD_COMMAND ${MAKE_PROGRAM} -j ${NUM_JOBS}
-  INSTALL_COMMAND ${HOST_ENV_CMAKE} ${MAKE_PROGRAM} install)
+  INSTALL_COMMAND ${HOST_ENV_CMAKE} ${MAKE_PROGRAM} install
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 add_library(ICU::ICU UNKNOWN IMPORTED)
 add_dependencies(ICU::ICU ICU-build)

CMake/resolve_dependency_modules/protobuf.cmake

@@ -39,7 +39,8 @@ FetchContent_Declare(
   protobuf
   URL ${VELOX_PROTOBUF_SOURCE_URL}
   URL_HASH ${VELOX_PROTOBUF_BUILD_SHA256_CHECKSUM}
-  OVERRIDE_FIND_PACKAGE EXCLUDE_FROM_ALL SYSTEM)
+  OVERRIDE_FIND_PACKAGE EXCLUDE_FROM_ALL SYSTEM CMAKE_ARGS
+  ${CMAKE_COMPILER_SELECTION_ARGS})
 
 set(protobuf_BUILD_TESTS OFF)
 set(protobuf_ABSL_PROVIDER

CMake/resolve_dependency_modules/pybind11.cmake

@@ -27,6 +27,7 @@ message(STATUS "Building Pybind11 from source")
 FetchContent_Declare(
   pybind11
   URL ${VELOX_PYBIND11_SOURCE_URL}
-  URL_HASH ${VELOX_PYBIND11_BUILD_SHA256_CHECKSUM})
+  URL_HASH ${VELOX_PYBIND11_BUILD_SHA256_CHECKSUM}
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 FetchContent_MakeAvailable(pybind11)

CMake/resolve_dependency_modules/re2.cmake

@@ -28,7 +28,8 @@ message(STATUS "Building re2 from source")
 FetchContent_Declare(
   re2
   URL ${VELOX_RE2_SOURCE_URL}
-  URL_HASH SHA256=${VELOX_RE2_BUILD_SHA256_CHECKSUM})
+  URL_HASH SHA256=${VELOX_RE2_BUILD_SHA256_CHECKSUM}
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 set(RE2_USE_ICU ON)
 set(RE2_BUILD_TESTING OFF)

CMake/resolve_dependency_modules/simdjson.cmake

@@ -27,6 +27,7 @@ message(STATUS "Building simdjson from source")
 FetchContent_Declare(
   simdjson
   URL ${VELOX_SIMDJSON_SOURCE_URL}
-  URL_HASH ${VELOX_SIMDJSON_BUILD_SHA256_CHECKSUM})
+  URL_HASH ${VELOX_SIMDJSON_BUILD_SHA256_CHECKSUM}
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 FetchContent_MakeAvailable(simdjson)

CMake/resolve_dependency_modules/stemmer.cmake

@@ -42,7 +42,7 @@ ExternalProject_Add(
   PATCH_COMMAND git apply ${CMAKE_CURRENT_LIST_DIR}/libstemmer/Makefile.patch
   BUILD_BYPRODUCTS
     ${STEMMER_PREFIX}/src/libstemmer/${CMAKE_STATIC_LIBRARY_PREFIX}stemmer${CMAKE_STATIC_LIBRARY_SUFFIX}
-)
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 add_library(stemmer STATIC IMPORTED GLOBAL)
 add_library(stemmer::stemmer ALIAS stemmer)

CMake/resolve_dependency_modules/template.cmake

@@ -25,6 +25,7 @@ message(STATUS "Building <PACKAGE> from source")
 FetchContent_Declare(
   <package>
   URL ${VELOX_<PACKAGE>_SOURCE_URL}
-  URL_HASH ${VELOX_<PACKAGE>_BUILD_SHA256_CHECKSUM})
+  URL_HASH ${VELOX_<PACKAGE>_BUILD_SHA256_CHECKSUM}
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 FetchContent_MakeAvailable(<package>)

CMake/resolve_dependency_modules/xsimd.cmake

@@ -26,6 +26,7 @@ message(STATUS "Building xsimd from source")
 FetchContent_Declare(
   xsimd
   URL ${VELOX_XSIMD_SOURCE_URL}
-  URL_HASH ${VELOX_XSIMD_BUILD_SHA256_CHECKSUM})
+  URL_HASH ${VELOX_XSIMD_BUILD_SHA256_CHECKSUM}
+  CMAKE_ARGS ${CMAKE_COMPILER_SELECTION_ARGS})
 
 FetchContent_MakeAvailable(xsimd)

CMakeLists.txt

@@ -206,6 +206,10 @@ if(${VELOX_FORCE_COLORED_OUTPUT})
   elseif("${CMAKE_CXX_COMPILER_ID}" STREQUAL "Clang"
          OR "${CMAKE_CXX_COMPILER_ID}" STREQUAL "AppleClang")
     add_compile_options(-fcolor-diagnostics)
+    if(CMAKE_SYSTEM_NAME STREQUAL "Linux" AND "${CMAKE_CXX_COMPILER_VERSION}"
+                                              VERSION_GREATER_EQUAL 15)
+      set(CMAKE_EXE_LINKER_FLAGS "-latomic")
+    endif()
   endif()
 endif()
 

Makefile

@@ -208,3 +208,14 @@ python-build:
 python-test:
 	$(MAKE) python-build extras="[tests]"
 	DEBUG=1 ${PYTHON_EXECUTABLE} -m unittest -v
+
+clang-debug:				#: Build with debugging symbols using Clang
+	$(MAKE) debug EXTRA_CMAKE_FLAGS=" ${EXTRA_CMAKE_FLAGS} \
+                                            -DCMAKE_C_COMPILER=clang \
+                                            -DCMAKE_CXX_COMPILER=clang++"
+
+
+clang-release:				#: Build the release version using Clang
+	$(MAKE) release EXTRA_CMAKE_FLAGS=" ${EXTRA_CMAKE_FLAGS} \
+                                            -DCMAKE_C_COMPILER=clang \
+                                            -DCMAKE_CXX_COMPILER=clang++"

scripts/setup-centos9.sh

@@ -31,18 +31,28 @@ set -efx -o pipefail
 SCRIPTDIR=$(dirname "${BASH_SOURCE[0]}")
 source $SCRIPTDIR/setup-helper-functions.sh
 NPROC=$(getconf _NPROCESSORS_ONLN)
-export CFLAGS=$(get_cxx_flags)  # Used by LZO.
-export CXXFLAGS=$CFLAGS  # Used by boost.
-export CPPFLAGS=$CFLAGS  # Used by LZO.
+export CXXFLAGS=$(get_cxx_flags) # Used by boost. 
+export CFLAGS=${CXXFLAGS//"-std=c++17"/} # Used by LZO.
 CMAKE_BUILD_TYPE="${BUILD_TYPE:-Release}"
 BUILD_DUCKDB="${BUILD_DUCKDB:-true}"
-export CC=/opt/rh/gcc-toolset-12/root/bin/gcc
-export CXX=/opt/rh/gcc-toolset-12/root/bin/g++
+USE_CLANG="${USE_CLANG:-false}"
 
 function dnf_install {
   dnf install -y -q --setopt=install_weak_deps=False "$@"
 }
 
+function install_and_setup_clang15 {
+  dnf_install clang15 gcc-toolset-13-libatomic-devel
+  CLANG_NUM_ALTERNATIVES=`alternatives --list | grep "clang " | wc -l`
+  if [[ ${CLANG_NUM_ALTERNATIVES} -eq 0 ]]; then
+    alternatives --install /usr/bin/clang clang /usr/bin/clang-15 80 --follower /usr/bin/clang++ clang++ /usr/bin/clang++-15
+  fi
+  (test ! -h /usr/local/bin/clang && ln -s /usr/bin/ccache /usr/local/bin/clang) || true
+  (test ! -h /usr/local/bin/clang++ && ln -s /usr/bin/ccache /usr/local/bin/clang++) || true
+  (test ! -h /usr/lib64/ccache/clang && ln -s /usr/bin/ccache /usr/lib64/ccache/clang) || true
+  (test ! -h /usr/lib64/ccache/clang++ && ln -s /usr/bin/ccache /usr/lib64/ccache/clang++) || true
+}
+
 # Install packages required for build.
 function install_build_prerequisites {
   dnf update -y
@@ -51,7 +61,12 @@ function install_build_prerequisites {
   dnf update -y
   dnf_install ninja-build cmake ccache gcc-toolset-12 git wget which
   dnf_install autoconf automake python3-devel pip libtool
+
   pip install cmake==3.28.3
+
+  if [[ ${USE_CLANG} != "false" ]]; then
+    install_and_setup_clang15
+  fi
 }
 
 # Install dependencies from the package managers.
@@ -100,9 +115,14 @@ function install_lzo {
 function install_boost {
   wget_and_untar https://github.com/boostorg/boost/releases/download/boost-1.84.0/boost-1.84.0.tar.gz boost
   (
-   cd boost
-   ./bootstrap.sh --prefix=/usr/local
-   ./b2 "-j$(nproc)" -d0 install threading=multi --without-python
+    cd boost
+    if [[ ${USE_CLANG} != "false" ]]; then
+      ./bootstrap.sh --prefix=/usr/local --with-toolset=clang
+      ./b2 "-j$(nproc)" -d0 install threading=multi toolset=clang --without-python 
+    else
+      ./bootstrap.sh --prefix=/usr/local
+      ./b2 "-j$(nproc)" -d0 install threading=multi --without-python
+    fi
   )
 }
 
@@ -244,9 +264,15 @@ function install_velox_deps {
 
 (
   if [[ $# -ne 0 ]]; then
-    # Activate gcc12; enable errors on unset variables afterwards.
-    source /opt/rh/gcc-toolset-12/enable || exit 1
-    set -u
+    if [[ ${USE_CLANG} != "false" ]]; then
+      export CC=/usr/bin/clang
+      export CXX=/usr/bin/clang++
+    else
+      # Activate gcc12; enable errors on unset variables afterwards.
+      source /opt/rh/gcc-toolset-12/enable || exit 1
+      set -u
+    fi
+
     for cmd in "$@"; do
       run_and_time "${cmd}"
     done
@@ -258,9 +284,14 @@ function install_velox_deps {
     else
       echo "Skipping installation of build dependencies since INSTALL_PREREQUISITES is not set"
     fi
-    # Activate gcc12; enable errors on unset variables afterwards.
-    source /opt/rh/gcc-toolset-12/enable || exit 1
-    set -u
+    if [[ ${USE_CLANG} != "false" ]]; then
+      export CC=/usr/bin/clang
+      export CXX=/usr/bin/clang++
+    else
+      # Activate gcc12; enable errors on unset variables afterwards.
+      source /opt/rh/gcc-toolset-12/enable || exit 1
+      set -u
+    fi
     install_velox_deps
     echo "All dependencies for Velox installed!"
     dnf clean all