Merge branch 'branch-24.12' into cudf-polars-string-numeric-casting

.devcontainer/cuda12.5-conda/devcontainer.json

@@ -15,9 +15,31 @@
   ],
   "hostRequirements": {"gpu": "optional"},
   "features": {
+    "ghcr.io/rapidsai/devcontainers/features/cuda:24.12": {
+        "version": "12.5",
+        "installCompilers": false,
+        "installProfilers": true,
+        "installDevPackages": false,
+        "installcuDNN": false,
+        "installcuTensor": false,
+        "installNCCL": false,
+        "installCUDARuntime": false,
+        "installNVRTC": false,
+        "installOpenCL": false,
+        "installcuBLAS": false,
+        "installcuSPARSE": false,
+        "installcuFFT": false,
+        "installcuFile": false,
+        "installcuRAND": false,
+        "installcuSOLVER": false,
+        "installNPP": false,
+        "installnvJPEG": false,
+        "pruneStaticLibs": true
+      },
     "ghcr.io/rapidsai/devcontainers/features/rapids-build-utils:24.12": {}
   },
   "overrideFeatureInstallOrder": [
+    "ghcr.io/rapidsai/devcontainers/features/cuda",
     "ghcr.io/rapidsai/devcontainers/features/rapids-build-utils"
   ],
   "initializeCommand": ["/bin/bash", "-c", "mkdir -m 0755 -p ${localWorkspaceFolder}/../.{aws,cache,config,conda/pkgs,conda/${localWorkspaceFolderBasename}-cuda12.5-envs}"],

.github/workflows/test.yaml

@@ -47,11 +47,23 @@ jobs:
     secrets: inherit
     uses: rapidsai/shared-workflows/.github/workflows/[email protected]
     with:
-      build_type: pull-request
+      build_type: nightly
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
       # Use the wheel container so we can skip conda solves and since our
       # primary static consumers (Spark) are not in conda anyway.
       container_image: "rapidsai/ci-wheel:latest"
       run_script: "ci/configure_cpp_static.sh"
+  clang-tidy:
+    secrets: inherit
+    uses: rapidsai/shared-workflows/.github/workflows/[email protected]
+    with:
+      build_type: nightly
+      branch: ${{ inputs.branch }}
+      date: ${{ inputs.date }}
+      sha: ${{ inputs.sha }}
+      run_script: "ci/clang_tidy.sh"
   conda-python-cudf-tests:
     secrets: inherit
     uses: rapidsai/shared-workflows/.github/workflows/[email protected]

ci/clang_tidy.sh

@@ -0,0 +1,29 @@
+#!/bin/bash
+# Copyright (c) 2024, NVIDIA CORPORATION.
+
+set -euo pipefail
+
+rapids-logger "Create clang-tidy conda environment"
+. /opt/conda/etc/profile.d/conda.sh
+
+ENV_YAML_DIR="$(mktemp -d)"
+
+rapids-dependency-file-generator \
+  --output conda \
+  --file-key clang_tidy \
+  --matrix "cuda=${RAPIDS_CUDA_VERSION%.*};arch=$(arch);py=${RAPIDS_PY_VERSION}" | tee "${ENV_YAML_DIR}/env.yaml"
+
+rapids-mamba-retry env create --yes -f "${ENV_YAML_DIR}/env.yaml" -n clang_tidy
+
+# Temporarily allow unbound variables for conda activation.
+set +u
+conda activate clang_tidy
+set -u
+
+RAPIDS_VERSION_MAJOR_MINOR="$(rapids-version-major-minor)"
+
+source rapids-configure-sccache
+
+# Run the build via CMake, which will run clang-tidy when CUDF_CLANG_TIDY is enabled.
+cmake -S cpp -B cpp/build -DCMAKE_BUILD_TYPE=Release -DCUDF_CLANG_TIDY=ON -GNinja
+cmake --build cpp/build

ci/release/update-version.sh

@@ -93,6 +93,7 @@ sed_runner "s/cudf-.*-SNAPSHOT/cudf-${NEXT_FULL_JAVA_TAG}/g" java/ci/README.md
 # .devcontainer files
 find .devcontainer/ -type f -name devcontainer.json -print0 | while IFS= read -r -d '' filename; do
     sed_runner "s@rapidsai/devcontainers:[0-9.]*@rapidsai/devcontainers:${NEXT_SHORT_TAG}@g" "${filename}"
+    sed_runner "s@rapidsai/devcontainers/features/cuda:[0-9.]*@rapidsai/devcontainers/features/cuda:${NEXT_SHORT_TAG_PEP440}@" "${filename}"
     sed_runner "s@rapidsai/devcontainers/features/rapids-build-utils:[0-9.]*@rapidsai/devcontainers/features/rapids-build-utils:${NEXT_SHORT_TAG_PEP440}@" "${filename}"
     sed_runner "s@rapids-\${localWorkspaceFolderBasename}-[0-9.]*@rapids-\${localWorkspaceFolderBasename}-${NEXT_SHORT_TAG}@g" "${filename}"
 done

cpp/.clang-tidy

@@ -39,7 +39,7 @@ Checks:
        -clang-analyzer-optin.core.EnumCastOutOfRange,
        -clang-analyzer-optin.cplusplus.UninitializedObject'
 
-WarningsAsErrors: ''
+WarningsAsErrors: '*'
 HeaderFilterRegex: '.*cudf/cpp/(src|include|tests).*'
 ExcludeHeaderFilterRegex: '.*(Message_generated.h|Schema_generated.h|brotli_dict.hpp|unbz2.hpp|cxxopts.hpp).*'
 FormatStyle:     none

cpp/CMakeLists.txt

@@ -88,6 +88,7 @@ option(
   ${DEFAULT_CUDF_BUILD_STREAMS_TEST_UTIL}
 )
 mark_as_advanced(CUDF_BUILD_STREAMS_TEST_UTIL)
+option(CUDF_CLANG_TIDY "Enable clang-tidy checking" OFF)
 
 message(VERBOSE "CUDF: Build with NVTX support: ${USE_NVTX}")
 message(VERBOSE "CUDF: Configure CMake to build tests: ${BUILD_TESTS}")
@@ -144,6 +145,58 @@ if(NOT CUDF_GENERATED_INCLUDE_DIR)
   set(CUDF_GENERATED_INCLUDE_DIR ${CUDF_BINARY_DIR})
 endif()
 
+# ##################################################################################################
+# * clang-tidy configuration ----------------------------------------------------------------------
+if(CUDF_CLANG_TIDY)
+  find_program(
+    CLANG_TIDY_EXE
+    NAMES "clang-tidy"
+    DOC "Path to clang-tidy executable" REQUIRED
+  )
+
+  execute_process(
+    COMMAND ${CLANG_TIDY_EXE} --version
+    OUTPUT_VARIABLE CLANG_TIDY_OUTPUT
+    OUTPUT_STRIP_TRAILING_WHITESPACE
+  )
+  string(REGEX MATCH "LLVM version ([0-9]+\\.[0-9]+)\\.[0-9]+" LLVM_VERSION_MATCH
+               "${CLANG_TIDY_OUTPUT}"
+  )
+  # Discard the patch version and allow it to float. Empirically, results between patch versions are
+  # mostly stable, and looking at available packages on some package managers sometimes patch
+  # versions are skipped so we don't want to constrain to a patch version that the user can't
+  # install.
+  set(LLVM_VERSION "${CMAKE_MATCH_1}")
+  set(expected_clang_tidy_version 19.1)
+  if(NOT expected_clang_tidy_version VERSION_EQUAL LLVM_VERSION)
+    message(
+      FATAL_ERROR
+        "clang-tidy version ${expected_clang_tidy_version} is required, but found ${LLVM_VERSION}"
+    )
+  endif()
+endif()
+
+# Turn on the clang-tidy property for a target excluding the files specified in SKIPPED_FILES.
+function(enable_clang_tidy target)
+  set(_tidy_options)
+  set(_tidy_one_value)
+  set(_tidy_multi_value SKIPPED_FILES)
+  cmake_parse_arguments(
+    _TIDY "${_tidy_options}" "${_tidy_one_value}" "${_tidy_multi_value}" ${ARGN}
+  )
+
+  if(CUDF_CLANG_TIDY)
+    # clang will complain about unused link libraries on the compile line unless we specify
+    # -Qunused-arguments.
+    set_target_properties(
+      ${target} PROPERTIES CXX_CLANG_TIDY "${CLANG_TIDY_EXE};--extra-arg=-Qunused-arguments"
+    )
+    foreach(file IN LISTS _TIDY_SKIPPED_FILES)
+      set_source_files_properties(${file} PROPERTIES SKIP_LINTING ON)
+    endforeach()
+  endif()
+endfunction()
+
 # ##################################################################################################
 # * conda environment -----------------------------------------------------------------------------
 rapids_cmake_support_conda_env(conda_env MODIFY_PREFIX_PATH)
@@ -714,6 +767,7 @@ target_compile_options(
   cudf PRIVATE "$<$<COMPILE_LANGUAGE:CXX>:${CUDF_CXX_FLAGS}>"
                "$<$<COMPILE_LANGUAGE:CUDA>:${CUDF_CUDA_FLAGS}>"
 )
+enable_clang_tidy(cudf SKIPPED_FILES src/io/comp/cpu_unbz2.cpp src/io/comp/brotli_dict.cpp)
 
 if(CUDF_BUILD_STACKTRACE_DEBUG)
   # Remove any optimization level to avoid nvcc warning "incompatible redefinition for option
@@ -863,15 +917,7 @@ if(CUDF_BUILD_TESTUTIL)
 
   add_library(cudf::cudftest_default_stream ALIAS cudftest_default_stream)
 
-  add_library(
-    cudftestutil SHARED
-    tests/io/metadata_utilities.cpp
-    tests/utilities/column_utilities.cu
-    tests/utilities/debug_utilities.cu
-    tests/utilities/random_seed.cpp
-    tests/utilities/table_utilities.cu
-    tests/utilities/tdigest_utilities.cu
-  )
+  add_library(cudftestutil INTERFACE)
 
   set_target_properties(
     cudftestutil
@@ -880,32 +926,56 @@ if(CUDF_BUILD_TESTUTIL)
                # set target compile options
                CXX_STANDARD 17
                CXX_STANDARD_REQUIRED ON
-               CXX_VISIBILITY_PRESET hidden
                CUDA_STANDARD 17
                CUDA_STANDARD_REQUIRED ON
-               CUDA_VISIBILITY_PRESET hidden
-               POSITION_INDEPENDENT_CODE ON
-               INTERFACE_POSITION_INDEPENDENT_CODE ON
   )
 
   target_compile_options(
-    cudftestutil PUBLIC "$<BUILD_INTERFACE:$<$<COMPILE_LANGUAGE:CXX>:${CUDF_CXX_FLAGS}>>"
-                        "$<BUILD_INTERFACE:$<$<COMPILE_LANGUAGE:CUDA>:${CUDF_CUDA_FLAGS}>>"
+    cudftestutil INTERFACE "$<BUILD_INTERFACE:$<$<COMPILE_LANGUAGE:CXX>:${CUDF_CXX_FLAGS}>>"
+                           "$<BUILD_INTERFACE:$<$<COMPILE_LANGUAGE:CUDA>:${CUDF_CUDA_FLAGS}>>"
   )
 
   target_link_libraries(
-    cudftestutil
-    PUBLIC Threads::Threads cudf cudftest_default_stream
-    PRIVATE GTest::gmock GTest::gtest $<TARGET_NAME_IF_EXISTS:conda_env>
+    cudftestutil INTERFACE Threads::Threads cudf cudftest_default_stream
+                           $<TARGET_NAME_IF_EXISTS:conda_env>
   )
 
   target_include_directories(
-    cudftestutil PUBLIC "$<BUILD_INTERFACE:${CUDF_SOURCE_DIR}>"
-                        "$<BUILD_INTERFACE:${CUDF_SOURCE_DIR}/src>"
+    cudftestutil INTERFACE "$<BUILD_INTERFACE:${CUDF_SOURCE_DIR}>"
+                           "$<BUILD_INTERFACE:${CUDF_SOURCE_DIR}/src>"
   )
   rapids_cuda_set_runtime(cudftestutil USE_STATIC ${CUDA_STATIC_RUNTIME})
   add_library(cudf::cudftestutil ALIAS cudftestutil)
 
+  add_library(cudftestutil_impl INTERFACE)
+  add_library(cudf::cudftestutil_impl ALIAS cudftestutil_impl)
+  target_sources(
+    cudftestutil_impl
+    INTERFACE $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/tests/io/metadata_utilities.cpp>
+              $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/tests/utilities/column_utilities.cu>
+              $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/tests/utilities/debug_utilities.cu>
+              $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/tests/utilities/random_seed.cpp>
+              $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/tests/utilities/table_utilities.cu>
+              $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/tests/utilities/tdigest_utilities.cu>
+              $<INSTALL_INTERFACE:src/cudftestutil/io/metadata_utilities.cpp>
+              $<INSTALL_INTERFACE:src/cudftestutil/utilities/column_utilities.cu>
+              $<INSTALL_INTERFACE:src/cudftestutil/utilities/debug_utilities.cu>
+              $<INSTALL_INTERFACE:src/cudftestutil/utilities/random_seed.cpp>
+              $<INSTALL_INTERFACE:src/cudftestutil/utilities/table_utilities.cu>
+              $<INSTALL_INTERFACE:src/cudftestutil/utilities/tdigest_utilities.cu>
+  )
+  target_link_libraries(cudftestutil_impl INTERFACE cudf::cudftestutil)
+
+  install(FILES tests/io/metadata_utilities.cpp DESTINATION src/cudftestutil/io)
+  install(
+    FILES tests/utilities/column_utilities.cu
+          tests/utilities/debug_utilities.cu
+          tests/utilities/random_seed.cpp
+          tests/utilities/table_utilities.cu
+          tests/utilities/tdigest_utilities.cu
+    DESTINATION src/cudftestutil/utilities
+  )
+
 endif()
 
 # * build cudf_identify_stream_usage --------------------------------------------------------------
@@ -1006,7 +1076,7 @@ install(
 set(_components_export_string)
 if(TARGET cudftestutil)
   install(
-    TARGETS cudftest_default_stream cudftestutil
+    TARGETS cudftest_default_stream cudftestutil cudftestutil_impl
     DESTINATION ${lib_dir}
     EXPORT cudf-testing-exports
   )
@@ -1046,14 +1116,15 @@ targets:
 This module offers an optional testing component which defines the
 following IMPORTED GLOBAL  targets:
 
- cudf::cudftestutil     - The main cudf testing library
+ cudf::cudftestutil          - The main cudf testing library
+ cudf::cudftestutil_impl     - C++ and CUDA sources to compile for definitions in cudf::cudftestutil
     ]=]
 )
 
 rapids_export(
   INSTALL cudf
   EXPORT_SET cudf-exports ${_components_export_string}
-  GLOBAL_TARGETS cudf cudftestutil
+  GLOBAL_TARGETS cudf cudftestutil cudftestutil_impl
   NAMESPACE cudf::
   DOCUMENTATION doc_string
 )
@@ -1074,7 +1145,7 @@ endif()
 rapids_export(
   BUILD cudf
   EXPORT_SET cudf-exports ${_components_export_string}
-  GLOBAL_TARGETS cudf cudftestutil
+  GLOBAL_TARGETS cudf cudftestutil cudftestutil_impl
   NAMESPACE cudf::
   DOCUMENTATION doc_string
   FINAL_CODE_BLOCK build_code_string

cpp/benchmarks/CMakeLists.txt

@@ -25,7 +25,7 @@ target_compile_options(
 target_link_libraries(
   cudf_datagen
   PUBLIC GTest::gmock GTest::gtest benchmark::benchmark nvbench::nvbench Threads::Threads cudf
-         cudftestutil nvtx3::nvtx3-cpp
+         cudf::cudftestutil nvtx3::nvtx3-cpp
   PRIVATE $<TARGET_NAME_IF_EXISTS:conda_env>
 )
 
@@ -49,7 +49,7 @@ target_compile_options(
 
 target_link_libraries(
   ndsh_data_generator
-  PUBLIC cudf cudftestutil nvtx3::nvtx3-cpp
+  PUBLIC cudf GTest::gmock GTest::gtest cudf::cudftestutil nvtx3::nvtx3-cpp
   PRIVATE $<TARGET_NAME_IF_EXISTS:conda_env>
 )
 
@@ -65,14 +65,14 @@ target_include_directories(
 # Use an OBJECT library so we only compile these helper source files only once
 add_library(
   cudf_benchmark_common OBJECT
-  "${CUDF_SOURCE_DIR}/tests/utilities/random_seed.cpp"
-  synchronization/synchronization.cpp
-  io/cuio_common.cpp
-  common/table_utilities.cpp
-  common/benchmark_utilities.cpp
-  common/nvbench_utilities.cpp
+  synchronization/synchronization.cpp io/cuio_common.cpp common/table_utilities.cpp
+  common/benchmark_utilities.cpp common/nvbench_utilities.cpp
 )
-target_link_libraries(cudf_benchmark_common PRIVATE cudf_datagen $<TARGET_NAME_IF_EXISTS:conda_env>)
+target_link_libraries(
+  cudf_benchmark_common PRIVATE cudf_datagen $<TARGET_NAME_IF_EXISTS:conda_env> GTest::gmock
+                                GTest::gtest
+)
+
 add_custom_command(
   OUTPUT CUDF_BENCHMARKS
   COMMAND echo Running benchmarks
@@ -99,7 +99,7 @@ function(ConfigureBench CMAKE_BENCH_NAME)
   )
   target_link_libraries(
     ${CMAKE_BENCH_NAME} PRIVATE cudf_benchmark_common cudf_datagen benchmark::benchmark_main
-                                $<TARGET_NAME_IF_EXISTS:conda_env>
+                                cudf::cudftestutil_impl $<TARGET_NAME_IF_EXISTS:conda_env>
   )
   add_custom_command(
     OUTPUT CUDF_BENCHMARKS
@@ -127,8 +127,9 @@ function(ConfigureNVBench CMAKE_BENCH_NAME)
                INSTALL_RPATH "\$ORIGIN/../../../lib"
   )
   target_link_libraries(
-    ${CMAKE_BENCH_NAME} PRIVATE cudf_benchmark_common ndsh_data_generator cudf_datagen
-                                nvbench::nvbench $<TARGET_NAME_IF_EXISTS:conda_env>
+    ${CMAKE_BENCH_NAME}
+    PRIVATE cudf_benchmark_common ndsh_data_generator cudf_datagen nvbench::nvbench
+            $<TARGET_NAME_IF_EXISTS:conda_env> cudf::cudftestutil_impl
   )
   install(
     TARGETS ${CMAKE_BENCH_NAME}