* update arrow to 18.1.0-2

2025-01-04 00:06:14 +01:00 · 2025-01-04 00:06:14 +01:00 · 0ae2608589
commit 0ae2608589
parent 4ceac269ab
3 changed files with 107 additions and 94 deletions
--- a/arrow/.SRCINFO
+++ b/arrow/.SRCINFO
@ -1,7 +1,7 @@
 pkgbase = arrow
 	pkgdesc = Columnar in-memory analytics layer for big data.
-	pkgver = 17.0.0
-	pkgrel = 9
+	pkgver = 18.1.0
+	pkgrel = 2
 	url = https://arrow.apache.org
 	arch = x86_64
 	arch = powerpc64le
@ -10,20 +10,21 @@ pkgbase = arrow
 	arch = riscv64
 	license = Apache-2.0
 	makedepends = boost
-	makedepends = clang
 	makedepends = cmake
 	makedepends = flatbuffers
 	makedepends = git
 	makedepends = gmock
-	makedepends = patchutils
+	makedepends = ninja
 	makedepends = rapidjson
 	makedepends = xsimd
+	depends = abseil-cpp
 	depends = apache-orc
 	depends = brotli
 	depends = bzip2
-	depends = gflags
-	depends = grpc
+	depends = gcc-libs
+	depends = glibc
 	depends = google-glog
+	depends = grpc
 	depends = libutf8proc
 	depends = lz4
 	depends = openssl
@ -35,28 +36,16 @@ pkgbase = arrow
 	depends = zstd
 	provides = parquet-cpp
 	conflicts = parquet-cpp
-	source = https://archive.apache.org/dist/arrow/arrow-17.0.0/apache-arrow-17.0.0.tar.gz
-	source = https://archive.apache.org/dist/arrow/arrow-17.0.0/apache-arrow-17.0.0.tar.gz.asc
-	source = https://github.com/apache/arrow/commit/58415d1fac50cb829b3dcf08526033d6db8c30db.patch
-	source = https://github.com/apache/arrow/commit/57ac40ca23ebcaa4d42ae808137033689d3be9b1.patch
-	source = https://github.com/apache/arrow/commit/8836535785ba3dd4ba335818a34e0479929b70e6.patch
-	source = https://github.com/apache/arrow/commit/69bce8f0cd02297ecc31caef22db67e654c16e28.patch
-	source = https://github.com/apache/arrow/commit/00d357674002b4e2e08b9d76b5d52530e723c4eb.patch
+	source = https://archive.apache.org/dist/arrow/arrow-18.1.0/apache-arrow-18.1.0.tar.gz
+	source = https://archive.apache.org/dist/arrow/arrow-18.1.0/apache-arrow-18.1.0.tar.gz.asc
 	source = git+https://github.com/apache/parquet-testing.git
 	source = git+https://github.com/apache/arrow-testing.git
-	source = lz4-cmake.patch
 	validpgpkeys = 265F80AB84FE03127E14F01125BCCA5220D84079
 	validpgpkeys = 08D3564B7C6A9CAFBFF6A66791D18FCF079F8007
 	validpgpkeys = AF6AADA4C9835B75973FF5DA275C532289DD0F4A
-	sha512sums = 4e2a617b8deeb9f94ee085653a721904a75696f0827bcba82b535cc7f4f723066a09914c7fa83c593e51a8a4031e8bf99e563cac1ebb1d89604cb406975d4864
-	sha512sums = SKIP
-	sha512sums = 533e985c350ddbd1b0666b460615a55f058919be4a6cf9b38777358fe9a91ef9bc8cc5c3b34128c75ccbe53bb31f713780f79799552f324093a9031b383ddeae
-	sha512sums = 8cdb5e3eb766786875dbd918ebf512dd834043dfb8ebff19f2fdad1a8ff81842b8b82cedd0e5ae1e87a08094ddda7f37035df805a439fc44febf42df40e023c4
-	sha512sums = 6cbbfd785bb613cfb87a83e73b65f9a40221b99aa20ad84e616cc410ed0073a621d9a37fa56772c4e6e89ffb6b13bd01950b78163f6b18122c3fac544a94a45d
-	sha512sums = 320d4863486cbcfb2f7c45d4f67dcf22502792ac2ac2f5b7d2713bd0b64d6f11a89f041e6341945eb0efb775b3d794fc25ce45a1b9d750d521a533f861295841
-	sha512sums = b4479a02f17585e4a8c8b809d22244720bc2f3be94463cd57c16e967aaa16348d3e3871ea64adfddd96392ffa967ec6822863efc29b5ab3ddf55f35b5807c626
+	sha512sums = 7249c03a6097bc64fb0092143e4d4aaef3227565147e6254f026ddd504177c8dd565a184a0df39743dc989070dc3785e5b66f738c8e310ed9c982b61c2ec4914
+	sha512sums = SKIP
 	sha512sums = SKIP
 	sha512sums = SKIP
-	sha512sums = 35af7cafbc83f02ab7fb219accda890ca1c233b0393bd2e91c07f3673a5a34e41752c5db071ab61fcc9e7154fafaca1394dd481de7e03e06856864c4197c52d6

 pkgname = arrow
--- a/arrow/.nvchecker.toml
+++ b/arrow/.nvchecker.toml
@ -0,0 +1,5 @@
+[arrow]
+source = "regex"
+url = "https://archive.apache.org/dist/arrow/"
+regex = "arrow-[\\d.]+"
+prefix = "arrow-"
--- a/arrow/PKGBUILD
+++ b/arrow/PKGBUILD
@ -1,107 +1,126 @@
 # POWER Maintainer: Alexander Baldeck <alex.bldck@gmail.com>
 # Maintainer: Bruno Pagani <archange@archlinux.org>
+# Maintainer: Jakub Klinkovský <lahwaacz at archlinux dot org>
 # Contributor: Guillaume Horel <guillaume.horel@gmail.com>

 pkgname=arrow
-pkgver=17.0.0
-pkgrel=9
+pkgver=18.1.0
+pkgrel=2
 pkgdesc="Columnar in-memory analytics layer for big data."
 arch=(x86_64 powerpc64le powerpc64 powerpc riscv64)
 url="https://arrow.apache.org"
 license=(Apache-2.0)
-depends=(apache-orc brotli bzip2 gflags grpc google-glog libutf8proc
-         lz4 openssl protobuf re2 snappy thrift zlib zstd)
+depends=(
+  abseil-cpp
+  apache-orc
+  brotli
+  bzip2
+  gcc-libs
+  glibc
+  google-glog
+  grpc
+  libutf8proc
+  lz4
+  openssl
+  protobuf
+  re2
+  snappy
+  thrift
+  zlib
+  zstd
+)
+makedepends=(
+  boost
+  cmake
+  flatbuffers
+  git
+  gmock
+  ninja
+  rapidjson
+  xsimd
+)
 provides=(parquet-cpp)
 conflicts=(parquet-cpp)
-makedepends=(boost clang cmake flatbuffers git gmock patchutils rapidjson xsimd)
-source=(https://archive.apache.org/dist/${pkgname}/${pkgname}-${pkgver}/apache-${pkgname}-${pkgver}.tar.gz{,.asc}
-        https://github.com/apache/arrow/commit/58415d1fac50cb829b3dcf08526033d6db8c30db.patch # [PATCH] GH-38183: [CI][Python] Use pipx to install GCS testbench (#43852)
-        https://github.com/apache/arrow/commit/57ac40ca23ebcaa4d42ae808137033689d3be9b1.patch # [PATCH] GH-43164: [C++] Fix CMake link order for AWS SDK (#43230)
-        https://github.com/apache/arrow/commit/8836535785ba3dd4ba335818a34e0479929b70e6.patch # [PATCH] GH-43702: [C++][FS][Azure] Use the latest Azurite and update the bundled Azure SDK for C++ to azure-identity_1.9.0 (#43723)
-        https://github.com/apache/arrow/commit/69bce8f0cd02297ecc31caef22db67e654c16e28.patch # [PATCH] GH-43677: [C++][FlightRPC] Move the FlightTestServer to its own .cc and .h files (#43678)
-        https://github.com/apache/arrow/commit/00d357674002b4e2e08b9d76b5d52530e723c4eb.patch # [PATCH] GH-43746: [C++] Add support for Boost 1.86 (#43766)
-        git+https://github.com/apache/parquet-testing.git
-        git+https://github.com/apache/arrow-testing.git
-        lz4-cmake.patch)
-sha512sums=('4e2a617b8deeb9f94ee085653a721904a75696f0827bcba82b535cc7f4f723066a09914c7fa83c593e51a8a4031e8bf99e563cac1ebb1d89604cb406975d4864'
-            'SKIP'
-            '533e985c350ddbd1b0666b460615a55f058919be4a6cf9b38777358fe9a91ef9bc8cc5c3b34128c75ccbe53bb31f713780f79799552f324093a9031b383ddeae'
-            '8cdb5e3eb766786875dbd918ebf512dd834043dfb8ebff19f2fdad1a8ff81842b8b82cedd0e5ae1e87a08094ddda7f37035df805a439fc44febf42df40e023c4'
-            '6cbbfd785bb613cfb87a83e73b65f9a40221b99aa20ad84e616cc410ed0073a621d9a37fa56772c4e6e89ffb6b13bd01950b78163f6b18122c3fac544a94a45d'
-            '320d4863486cbcfb2f7c45d4f67dcf22502792ac2ac2f5b7d2713bd0b64d6f11a89f041e6341945eb0efb775b3d794fc25ce45a1b9d750d521a533f861295841'
-            'b4479a02f17585e4a8c8b809d22244720bc2f3be94463cd57c16e967aaa16348d3e3871ea64adfddd96392ffa967ec6822863efc29b5ab3ddf55f35b5807c626'
+source=(
+  https://archive.apache.org/dist/$pkgname/$pkgname-$pkgver/apache-$pkgname-$pkgver.tar.gz{,.asc}
+  git+https://github.com/apache/parquet-testing.git
+  git+https://github.com/apache/arrow-testing.git
+)
+sha512sums=('7249c03a6097bc64fb0092143e4d4aaef3227565147e6254f026ddd504177c8dd565a184a0df39743dc989070dc3785e5b66f738c8e310ed9c982b61c2ec4914'
            'SKIP'
            'SKIP'
-            '35af7cafbc83f02ab7fb219accda890ca1c233b0393bd2e91c07f3673a5a34e41752c5db071ab61fcc9e7154fafaca1394dd481de7e03e06856864c4197c52d6')
+            'SKIP')
 validpgpkeys=(265F80AB84FE03127E14F01125BCCA5220D84079  # Krisztian Szucs (apache) <szucs.krisztian@gmail.com>
              08D3564B7C6A9CAFBFF6A66791D18FCF079F8007  # Kouhei Sutou <kou@cozmixng.org>
              AF6AADA4C9835B75973FF5DA275C532289DD0F4A) # Raúl Cumplido Domínguez (CODE SIGNING KEY) <raulcd@apache.org>

-prepare() {
-  patch -d apache-${pkgname}-${pkgver} -p1 < lz4-cmake.patch # Fix build when lz4 is built with cmake
-  filterdiff -x .github/workflows/cpp.yml -x 'ci/*' -p1 58415d1fac50cb829b3dcf08526033d6db8c30db.patch \
-    | patch -d apache-${pkgname}-${pkgver} -p1
-  patch -d apache-${pkgname}-${pkgver} -p1 < 57ac40ca23ebcaa4d42ae808137033689d3be9b1.patch
-  patch -d apache-${pkgname}-${pkgver} -p1 < 8836535785ba3dd4ba335818a34e0479929b70e6.patch
-  patch -d apache-${pkgname}-${pkgver} -p1 < 69bce8f0cd02297ecc31caef22db67e654c16e28.patch
-  patch -d apache-${pkgname}-${pkgver} -p1 < 00d357674002b4e2e08b9d76b5d52530e723c4eb.patch
-}
+build() {
+  # Arrow options are defined in https://github.com/apache/arrow/blob/main/cpp/cmake_modules/DefineOptions.cmake
+  # Documentation for building from source: https://arrow.apache.org/docs/developers/cpp/building.html
+  local cmake_options=(
+    -B build
+    -S apache-$pkgname-$pkgver/cpp
+    -G Ninja
+    -W no-dev
+    # one pyarrow test fails when the build type is none of 'debug', 'release', 'minsizerel', 'relwithdebinfo'
+    -DCMAKE_BUILD_TYPE=RelWithDebInfo
+    -DCMAKE_INSTALL_PREFIX="/usr"
+    -DCMAKE_INSTALL_LIBDIR="lib"
+    -DARROW_SIMD_LEVEL=NONE
+    -DARROW_RUNTIME_SIMD_LEVEL=MAX
+    -DARROW_DEPENDENCY_SOURCE=SYSTEM
+    -DARROW_BUILD_STATIC=OFF
+    -DARROW_BUILD_UTILITIES=ON
+    -DARROW_COMPUTE=ON
+    -DARROW_CSV=ON
+    -DARROW_FILESYSTEM=ON
+    -DARROW_FLIGHT=ON
+    -DARROW_FLIGHT_SQL=ON
+    -DARROW_GANDIVA=OFF  # depends on LLVMAlt
+    -DARROW_HDFS=ON
+    -DARROW_IPC=ON
+    -DARROW_JEMALLOC=ON
+    -DARROW_JSON=ON
+    -DARROW_ORC=ON
+    -DARROW_PARQUET=ON
+    -DARROW_SUBSTRAIT=ON
+    -DARROW_TENSORFLOW=ON
+    -DARROW_USE_GLOG=ON
+    -DARROW_WITH_BROTLI=ON
+    -DARROW_WITH_BZ2=ON
+    -DARROW_WITH_LZ4=ON
+    -DARROW_WITH_SNAPPY=ON
+    -DARROW_WITH_ZLIB=ON
+    -DARROW_WITH_ZSTD=ON
+    -DPARQUET_REQUIRE_ENCRYPTION=ON
+  )

-build(){
  BUILD_TESTS='ON'
  case "${CARCH}" in
-    powerpc64le) _cmake_options+=(-DARROW_SIMD_LEVEL=NONE) ;;
+    powerpc64le) cmake_options+=(-DARROW_BUILD_TESTS=ON) ;;
    powerpc64)
-      _cmake_options+=(-DARROW_SIMD_LEVEL=NONE);
-      BUILD_TESTS='OFF'
+      cmake_options+=(-DARROW_SIMD_LEVEL=NONE -DARROW_BUILD_TESTS=OFF);
    ;;
-    powerpc|riscv64) _cmake_options+=(-DCMAKE_CXX_FLAGS="-latomic" -DARROW_SIMD_LEVEL=NONE) ;;
+    powerpc|riscv64) cmake_options+=(-DCMAKE_CXX_FLAGS="-latomic" -DARROW_SIMD_LEVEL=NONE -DARROW_BUILD_TESTS=ON) ;;
    x86_64)
-      _cmake_options+=(-DARROW_SIMD_LEVEL=AVX2)
+      _cmake_options+=(-DARROW_SIMD_LEVEL=AVX2 -DARROW_BUILD_TESTS=ON)
      export CC=clang
      export CXX=clang++
    ;;
  esac

-  cmake \
-    -B build -S apache-${pkgname}-${pkgver}/cpp \
-    -DCMAKE_INSTALL_PREFIX="/usr" \
-    -DCMAKE_INSTALL_LIBDIR="lib" \
-    -DCMAKE_BUILD_TYPE=Release \
-    -DARROW_BUILD_STATIC=OFF \
-    -DARROW_DEPENDENCY_SOURCE=SYSTEM \
-    -DARROW_BUILD_TESTS=${BUILD_TESTS} \
-    -DARROW_COMPUTE=ON \
-    -DARROW_CSV=ON \
-    -DARROW_SUBSTRAIT=ON \
-    -DARROW_FLIGHT=ON \
-    -DARROW_FLIGHT_SQL=ON \
-    -DARROW_GANDIVA=OFF \
-    -DARROW_HDFS=ON \
-    -DARROW_IPC=ON \
-    -DARROW_JEMALLOC=ON \
-    -DARROW_ORC=ON \
-    -DARROW_PARQUET=ON \
-    -DARROW_TENSORFLOW=ON \
-    -DARROW_USE_GLOG=ON \
-    -DARROW_WITH_BROTLI=ON \
-    -DARROW_WITH_BZ2=ON \
-    -DARROW_WITH_LZ4=ON \
-    -DARROW_WITH_SNAPPY=ON \
-    -DARROW_WITH_ZLIB=ON \
-    -DARROW_WITH_ZSTD=ON \
-    -DPARQUET_REQUIRE_ENCRYPTION=ON \
-    -Wno-dev ${_cmake_options[@]}
-  make -C build
+  cmake "${cmake_options[@]}"
+  cmake --build build
 }

-check(){
-  PARQUET_TEST_DATA="${srcdir}"/parquet-testing/data \
-  ARROW_TEST_DATA="${srcdir}"/arrow-testing/data \
+check() {
+  PARQUET_TEST_DATA="$srcdir"/parquet-testing/data \
+  ARROW_TEST_DATA="$srcdir"/arrow-testing/data \
  ctest --test-dir build --output-on-failure
 }

-package(){
+package() {
  DESTDIR="$pkgdir" cmake --install build
-  find "${pkgdir}"/usr/lib/ -name '*testing*' -delete
+  # remove testing libraries
+  find "$pkgdir"/usr/lib/ -name '*testing*' -delete
 }