[R-package] make package installable with CRAN toolchain (fixes #2960)

.ci/test_r_package.sh

@@ -50,10 +50,25 @@ if [[ $AZURE != "true" ]] && [[ $OS_NAME == "linux" ]]; then
             texlive-fonts-extra \
             qpdf \
             || exit -1
+
+    # https://github.com/r-lib/actions/issues/111
+    if [[ $R_BUILD_TYPE == "cran" ]]; then
+        sudo apt-get install \
+            --no-install-recommends \
+            -y \
+                autoconf=$(cat R-package/AUTOCONF_UBUNTU_VERSION) \
+                devscripts \
+                || exit -1
+    fi
 fi
 
 # Installing R precompiled for Mac OS 10.11 or higher
 if [[ $OS_NAME == "macos" ]]; then
+    if [[ $R_BUILD_TYPE == "cran" ]]; then
+        brew install \
+            automake \
+            checkbashisms
+    fi
     brew install qpdf
     brew cask install basictex
     export PATH="/Library/TeX/texbin:$PATH"
@@ -109,10 +124,35 @@ if [[ $TASK == "r-package-check-docs" ]]; then
 fi
 
 cd ${BUILD_DIRECTORY}
-Rscript build_r.R --skip-install || exit -1
 
 PKG_TARBALL="lightgbm_${LGB_VER}.tar.gz"
 LOG_FILE_NAME="lightgbm.Rcheck/00check.log"
+if [[ $R_BUILD_TYPE == "cmake" ]]; then
+    Rscript build_r.R --skip-install || exit -1
+elif [[ $R_BUILD_TYPE == "cran" ]]; then
+    ./build-cran-package.sh || exit -1
+
+    num_files_changed=$(
+        git diff --name-only | wc -l
+    )
+    if [[ ${num_files_changed} -gt 0 ]]; then
+        echo "'configure' in the R package has changed. Please recreate it and commit the changes."
+        echo "changed files:"
+        git diff --compact-summary
+        echo "See R-package/README.md for details on how to recreate this script."
+        echo ""
+        exit -1
+    fi
+
+    # Test CRAN source .tar.gz in a directory that is not this repo or below it.
+    # When people install.packages('lightgbm'), they won't have the LightGBM
+    # git repo around. This is to protect against the use of relative paths
+    # like ../../CMakeLists.txt that would only work if you are in the repo
+    R_CMD_CHECK_DIR="${HOME}/tmp-r-cmd-check/"
+    mkdir -p ${R_CMD_CHECK_DIR}
+    mv ${PKG_TARBALL} ${R_CMD_CHECK_DIR}
+    cd ${R_CMD_CHECK_DIR}
+fi
 
 # fails tests if either ERRORs or WARNINGs are thrown by
 # R CMD CHECK
@@ -134,7 +174,8 @@ while kill -0 ${CHECK_PID} >/dev/null 2>&1; do
 done
 
 echo "R CMD check build logs:"
-cat ${BUILD_DIRECTORY}/lightgbm.Rcheck/00install.out
+BUILD_LOG_FILE=lightgbm.Rcheck/00install.out
+cat ${BUILD_LOG_FILE}
 
 if [[ $check_succeeded == "no" ]]; then
     exit -1
@@ -145,7 +186,11 @@ if grep -q -R "WARNING" "$LOG_FILE_NAME"; then
     exit -1
 fi
 
-ALLOWED_CHECK_NOTES=2
+if [[ $OS_NAME == "linux" ]] && [[ $R_BUILD_TYPE == "cran" ]]; then
+    ALLOWED_CHECK_NOTES=2
+else
+    ALLOWED_CHECK_NOTES=1
+fi
 NUM_CHECK_NOTES=$(
     cat ${LOG_FILE_NAME} \
         | grep -e '^Status: .* NOTE.*' \
@@ -155,3 +200,17 @@ if [[ ${NUM_CHECK_NOTES} -gt ${ALLOWED_CHECK_NOTES} ]]; then
     echo "Found ${NUM_CHECK_NOTES} NOTEs from R CMD check. Only ${ALLOWED_CHECK_NOTES} are allowed"
     exit -1
 fi
+
+# this check makes sure that CI builds of the CRAN package on Mac
+# actually use OpenMP
+if [[ $OS_NAME == "macos" ]] && [[ $R_BUILD_TYPE == "cran" ]]; then
+    omp_working=$(
+        cat $BUILD_LOG_FILE \
+        | grep -E "checking whether OpenMP will work .*yes" \
+        | wc -l
+    )
+    if [[ $omp_working -ne 1 ]]; then
+        echo "OpenMP was not found, and should be when testing the CRAN package on macOS"
+        exit -1
+    fi
+fi

.ci/test_r_package_windows.ps1

@@ -44,54 +44,61 @@ function Run-R-Code-Redirect-Stderr {
   Rscript --vanilla -e $decorated_code
 }
 
-$env:R_LIB_PATH = "$env:BUILD_SOURCESDIRECTORY/RLibrary" -replace '[\\]', '/'
-$env:R_LIBS = "$env:R_LIB_PATH"
-$env:PATH = "$env:R_LIB_PATH/Rtools/bin;" + "$env:R_LIB_PATH/Rtools/usr/bin;" + "$env:R_LIB_PATH/R/bin/x64;" + "$env:R_LIB_PATH/miktex/texmfs/install/miktex/bin/x64;" + $env:PATH
-$env:CRAN_MIRROR = "https://cloud.r-project.org/"
-$env:CTAN_MIRROR = "https://ctan.math.illinois.edu/systems/win32/miktex"
-$env:CTAN_MIKTEX_ARCHIVE = "$env:CTAN_MIRROR/setup/windows-x64/"
-$env:CTAN_PACKAGE_ARCHIVE = "$env:CTAN_MIRROR/tm/packages/"
-
 # Get details needed for installing R components
 #
 # NOTES:
 #    * some paths and file names are different on R4.0
 $env:R_MAJOR_VERSION = $env:R_VERSION.split('.')[0]
 if ($env:R_MAJOR_VERSION -eq "3") {
-  $env:RTOOLS_MINGW_BIN = "$env:R_LIB_PATH/Rtools/mingw_64/bin"
+  # Rtools 3.x has to be installed at C:\Rtools\
+  #     * https://stackoverflow.com/a/46619260/3986677
+  $RTOOLS_INSTALL_PATH = "C:\Rtools"
+  $env:RTOOLS_MINGW_BIN = "$RTOOLS_INSTALL_PATH/mingw_64/bin"
   $env:RTOOLS_EXE_FILE = "Rtools35.exe"
   $env:R_WINDOWS_VERSION = "3.6.3"
 } elseif ($env:R_MAJOR_VERSION -eq "4") {
-  $env:RTOOLS_MINGW_BIN = "$env:R_LIB_PATH/Rtools/mingw64/bin"
+  $RTOOLS_INSTALL_PATH = "C:\rtools40"
+  $env:RTOOLS_MINGW_BIN = "$RTOOLS_INSTALL_PATH/mingw64/bin"
   $env:RTOOLS_EXE_FILE = "rtools40-x86_64.exe"
   $env:R_WINDOWS_VERSION = "4.0.2"
 } else {
   Write-Output "[ERROR] Unrecognized R version: $env:R_VERSION"
   Check-Output $false
 }
 
-if ($env:COMPILER -eq "MINGW") {
+$env:R_LIB_PATH = "$env:BUILD_SOURCESDIRECTORY/RLibrary" -replace '[\\]', '/'
+$env:R_LIBS = "$env:R_LIB_PATH"
+$env:PATH = "$RTOOLS_INSTALL_PATH/bin;" + "$RTOOLS_INSTALL_PATH/usr/bin;" + "$env:R_LIB_PATH/R/bin/x64;" + "$env:R_LIB_PATH/miktex/texmfs/install/miktex/bin/x64;" + $env:PATH
+$env:CRAN_MIRROR = "https://cloud.r-project.org/"
+$env:CTAN_MIRROR = "https://ctan.math.illinois.edu/systems/win32/miktex"
+$env:CTAN_MIKTEX_ARCHIVE = "$env:CTAN_MIRROR/setup/windows-x64/"
+$env:CTAN_PACKAGE_ARCHIVE = "$env:CTAN_MIRROR/tm/packages/"
+
+if (($env:COMPILER -eq "MINGW") -and ($env:R_BUILD_TYPE -eq "cmake")) {
   $env:CXX = "$env:RTOOLS_MINGW_BIN/g++.exe"
   $env:CC = "$env:RTOOLS_MINGW_BIN/gcc.exe"
 }
 
 cd $env:BUILD_SOURCESDIRECTORY
 tzutil /s "GMT Standard Time"
 [Void][System.IO.Directory]::CreateDirectory($env:R_LIB_PATH)
-
-if ($env:TOOLCHAIN -eq "MINGW") {
-  Write-Output "Telling R to use MinGW"
-  $install_libs = "$env:BUILD_SOURCESDIRECTORY/R-package/src/install.libs.R"
-  ((Get-Content -Path $install_libs -Raw) -Replace 'use_mingw <- FALSE','use_mingw <- TRUE') | Set-Content -Path $install_libs
-} elseif ($env:TOOLCHAIN -eq "MSYS") {
-  Write-Output "Telling R to use MSYS"
-  $install_libs = "$env:BUILD_SOURCESDIRECTORY/R-package/src/install.libs.R"
-  ((Get-Content -Path $install_libs -Raw) -Replace 'use_msys2 <- FALSE','use_msys2 <- TRUE') | Set-Content -Path $install_libs
-} elseif ($env:TOOLCHAIN -eq "MSVC") {
-  # no customization for MSVC
-} else {
-  Write-Output "[ERROR] Unrecognized compiler: $env:TOOLCHAIN"
-  Check-Output $false
+$env:LGB_VER = Get-Content -Path VERSION.txt -TotalCount 1
+
+if ($env:R_BUILD_TYPE -eq "cmake") {
+  if ($env:TOOLCHAIN -eq "MINGW") {
+    Write-Output "Telling R to use MinGW"
+    $install_libs = "$env:BUILD_SOURCESDIRECTORY/R-package/src/install.libs.R"
+    ((Get-Content -Path $install_libs -Raw) -Replace 'use_mingw <- FALSE','use_mingw <- TRUE') | Set-Content -Path $install_libs
+  } elseif ($env:TOOLCHAIN -eq "MSYS") {
+    Write-Output "Telling R to use MSYS"
+    $install_libs = "$env:BUILD_SOURCESDIRECTORY/R-package/src/install.libs.R"
+    ((Get-Content -Path $install_libs -Raw) -Replace 'use_msys2 <- FALSE','use_msys2 <- TRUE') | Set-Content -Path $install_libs
+  } elseif ($env:TOOLCHAIN -eq "MSVC") {
+    # no customization for MSVC
+  } else {
+    Write-Output "[ERROR] Unrecognized toolchain: $env:TOOLCHAIN"
+    Check-Output $false
+  }
 }
 
 # download R and RTools
@@ -105,16 +112,18 @@ Start-Process -FilePath R-win.exe -NoNewWindow -Wait -ArgumentList "/VERYSILENT
 Write-Output "Done installing R"
 
 Write-Output "Installing Rtools"
-Start-Process -FilePath Rtools.exe -NoNewWindow -Wait -ArgumentList "/VERYSILENT /DIR=$env:R_LIB_PATH/Rtools" ; Check-Output $?
+Start-Process -FilePath Rtools.exe -NoNewWindow -Wait -ArgumentList "/VERYSILENT /DIR=$RTOOLS_INSTALL_PATH" ; Check-Output $?
 Write-Output "Done installing Rtools"
 
 Write-Output "Installing dependencies"
 $packages = "c('data.table', 'jsonlite', 'Matrix', 'processx', 'R6', 'testthat'), dependencies = c('Imports', 'Depends', 'LinkingTo')"
 Run-R-Code-Redirect-Stderr "options(install.packages.check.source = 'no'); install.packages($packages, repos = '$env:CRAN_MIRROR', type = 'binary', lib = '$env:R_LIB_PATH')" ; Check-Output $?
 
-# MiKTeX and pandoc can be skipped on MSVC builds, since we don't
-# build the package documentation for those
-if ($env:COMPILER -ne "MSVC") {
+# MiKTeX and pandoc can be skipped on non-MinGW builds, since we don't
+# build the package documentation for those.
+#
+# MiKTeX always needs to be built to test a CRAN package.
+if (($env:COMPILER -eq "MINGW") -or ($env:R_BUILD_TYPE -eq "cran")) {
     Download-Miktex-Setup "$env:CTAN_MIKTEX_ARCHIVE" "miktexsetup-x64.zip"
     Add-Type -AssemblyName System.IO.Compression.FileSystem
     [System.IO.Compression.ZipFile]::ExtractToDirectory("miktexsetup-x64.zip", "miktex")
@@ -132,17 +141,29 @@ Write-Output "Building R package"
 
 # R CMD check is not used for MSVC builds
 if ($env:COMPILER -ne "MSVC") {
-  Run-R-Code-Redirect-Stderr "commandArgs <- function(...){'--skip-install'}; source('build_r.R')"; Check-Output $?
 
-  $PKG_FILE_NAME = Get-Item *.tar.gz
-  $PKG_FILE_NAME = $PKG_FILE_NAME -replace '[\\]', '/'
+  $PKG_FILE_NAME = "lightgbm_$env:LGB_VER.tar.gz"
   $LOG_FILE_NAME = "lightgbm.Rcheck/00check.log"
 
+  if ($env:R_BUILD_TYPE -eq "cmake") {
+    Run-R-Code-Redirect-Stderr "commandArgs <- function(...){'--skip-install'}; source('build_r.R')"; Check-Output $?
+  } elseif ($env:R_BUILD_TYPE -eq "cran") {
+    Run-R-Code-Redirect-Stderr "result <- processx::run(command = 'sh', args = 'build-cran-package.sh', echo = TRUE, windows_verbatim_args = FALSE)" ; Check-Output $?
+    # Test CRAN source .tar.gz in a directory that is not this repo or below it.
+    # When people install.packages('lightgbm'), they won't have the LightGBM
+    # git repo around. This is to protect against the use of relative paths
+    # like ../../CMakeLists.txt that would only work if you are in the repoo
+    $R_CMD_CHECK_DIR = "tmp-r-cmd-check"
+    New-Item -Path "C:\" -Name $R_CMD_CHECK_DIR -ItemType "directory" > $null
+    Move-Item -Path "$PKG_FILE_NAME" -Destination "C:\$R_CMD_CHECK_DIR\" > $null
+    cd "C:\$R_CMD_CHECK_DIR\"
+  }
+
   Write-Output "Running R CMD check as CRAN"
   Run-R-Code-Redirect-Stderr "result <- processx::run(command = 'R.exe', args = c('CMD', 'check', '--no-multiarch', '--as-cran', '$PKG_FILE_NAME'), echo = TRUE, windows_verbatim_args = FALSE)" ; $check_succeeded = $?
 
   Write-Output "R CMD check build logs:"
-  $INSTALL_LOG_FILE_NAME = "$env:BUILD_SOURCESDIRECTORY\lightgbm.Rcheck\00install.out"
+  $INSTALL_LOG_FILE_NAME = "lightgbm.Rcheck\00install.out"
   Get-Content -Path "$INSTALL_LOG_FILE_NAME"
 
   Check-Output $check_succeeded
@@ -175,14 +196,17 @@ if ($env:COMPILER -ne "MSVC") {
 # to fail back to MinGW if MSVC fails, but for CI builds we need to check that the correct
 # compiler was used.
 $checks = Select-String -Path "${INSTALL_LOG_FILE_NAME}" -Pattern "Check for working CXX compiler.*$env:COMPILER"
-if ($checks.Matches.length -eq 0) {
-  Write-Output "The wrong compiler was used. Check the build logs."
-  Check-Output $False
+if ($env:R_BUILD_TYPE -eq "cmake") {
+  $checks = Select-String -Path "${INSTALL_LOG_FILE_NAME}" -Pattern "Check for working CXX compiler.*$env:COMPILER"
+  if ($checks.Matches.length -eq 0) {
+    Write-Output "The wrong compiler was used. Check the build logs."
+    Check-Output $False
+  }
 }
 
 # Checking that we got the right toolchain for MinGW. If using MinGW, both
 # MinGW and MSYS toolchains are supported
-if ($env:COMPILER -eq "MINGW") {
+if (($env:COMPILER -eq "MINGW") -and ($env:R_BUILD_TYPE -eq "cmake")) {
   $checks = Select-String -Path "${INSTALL_LOG_FILE_NAME}" -Pattern "Trying to build with.*$env:TOOLCHAIN"
   if ($checks.Matches.length -eq 0) {
     Write-Output "The wrong toolchain was used. Check the build logs."

.github/workflows/main.yml

@@ -10,71 +10,110 @@ on:
 
 jobs:
   test:
-    name: ${{ matrix.task }} (${{ matrix.os }}, ${{ matrix.compiler }}, R ${{ matrix.r_version }})
+    name: ${{ matrix.task }} (${{ matrix.os }}, ${{ matrix.compiler }}, R ${{ matrix.r_version }}, ${{ matrix.build_type }})
     runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
         include:
+          ################
+          # CMake builds #
+          ################
           - os: ubuntu-latest
             task: r-package
             compiler: gcc
             r_version: 3.6
+            build_type: cmake
           - os: ubuntu-latest
             task: r-package
             compiler: gcc
             r_version: 4.0
+            build_type: cmake
           - os: ubuntu-latest
             task: r-package
             compiler: clang
             r_version: 3.6
+            build_type: cmake
           - os: ubuntu-latest
             task: r-package
             compiler: clang
             r_version: 4.0
+            build_type: cmake
           - os: ubuntu-latest
             task: r-package-check-docs
             compiler: gcc
             r_version: 4.0
+            build_type: cmake
           - os: macOS-latest
             task: r-package
             compiler: gcc
             r_version: 3.6
+            build_type: cmake
           - os: macOS-latest
             task: r-package
             compiler: gcc
             r_version: 4.0
+            build_type: cmake
           - os: macOS-latest
             task: r-package
             compiler: clang
             r_version: 3.6
+            build_type: cmake
           - os: macOS-latest
             task: r-package
             compiler: clang
             r_version: 4.0
+            build_type: cmake
           - os: windows-latest
             task: r-package
             compiler: MINGW
             toolchain: MINGW
             r_version: 3.6
+            build_type: cmake
           - os: windows-latest
             task: r-package
             compiler: MINGW
             toolchain: MSYS
             r_version: 4.0
+            build_type: cmake
           # Visual Studio 2017
           - os: windows-2016
             task: r-package
             compiler: MSVC
             toolchain: MSVC
             r_version: 3.6
+            build_type: cmake
           # Visual Studio 2019
           - os: windows-2019
             task: r-package
             compiler: MSVC
             toolchain: MSVC
             r_version: 4.0
+            build_type: cmake
+          ###############
+          # CRAN builds #
+          ###############
+          - os: windows-latest
+            task: r-package
+            compiler: MINGW
+            toolchain: MSYS
+            r_version: 4.0
+            build_type: cran
+          - os: ubuntu-latest
+            task: r-package
+            compiler: gcc
+            r_version: 4.0
+            build_type: cran
+          - os: macOS-latest
+            task: r-package
+            compiler: clang
+            r_version: 4.0
+            build_type: cran
     steps:
+      - name: Prevent conversion of line endings on Windows
+        if: startsWith(matrix.os, 'windows')
+        shell: pwsh
+        run: git config --global core.autocrlf false
       - name: Checkout repository
         uses: actions/checkout@v1
         with:
@@ -100,6 +139,7 @@ jobs:
           export PATH="$CONDA/bin:${HOME}/.local/bin:$PATH"
           export LGB_VER=$(head -n 1 VERSION.txt)
           export R_VERSION="${{ matrix.r_version }}"
+          export R_BUILD_TYPE="${{ matrix.build_type }}"
           $GITHUB_WORKSPACE/.ci/setup.sh
           $GITHUB_WORKSPACE/.ci/test.sh
       - name: Use conda on Windows
@@ -114,6 +154,7 @@ jobs:
           $env:BUILD_SOURCESDIRECTORY = $env:GITHUB_WORKSPACE
           $env:TOOLCHAIN = "${{ matrix.toolchain }}"
           $env:R_VERSION = "${{ matrix.r_version }}"
+          $env:R_BUILD_TYPE = "${{ matrix.build_type }}"
           $env:COMPILER = "${{ matrix.compiler }}"
           $env:GITHUB_ACTIONS = "true"
           $env:TASK = "${{ matrix.task }}"