Merge pull request #120854 from tobim/pkgs/arrow-cpp-4.0.0

arrow-cpp: 3.0.0 -> 4.0.0

pkgs/development/libraries/arrow-cpp/default.nix

@@ -1,6 +1,7 @@
-{ stdenv, lib, fetchurl, fetchFromGitHub, fetchpatch, fixDarwinDylibNames
+{ stdenv, lib, fetchurl, fetchFromGitHub, fixDarwinDylibNames
 , autoconf, boost, brotli, cmake, flatbuffers, gflags, glog, gtest, lz4
-, perl, python3, rapidjson, re2, snappy, thrift, utf8proc, which, zlib, zstd
+, perl, python3, rapidjson, re2, snappy, thrift, utf8proc, which, xsimd
+, zlib, zstd
 , enableShared ? !stdenv.hostPlatform.isStatic
 }:
 
@@ -15,18 +16,18 @@ let
   parquet-testing = fetchFromGitHub {
     owner = "apache";
     repo = "parquet-testing";
-    rev = "e31fe1a02c9e9f271e4bfb8002d403c52f1ef8eb";
-    sha256 = "02f51dvx8w5mw0bx3hn70hkn55mn1m65kzdps1ifvga9hghpy0sh";
+    rev = "ddd898958803cb89b7156c6350584d1cda0fe8de";
+    sha256 = "0n16xqlpxn2ryp43w8pppxrbwmllx6sk4hv3ycgikfj57nd3ibc0";
   };
 
 in stdenv.mkDerivation rec {
   pname = "arrow-cpp";
-  version = "3.0.0";
+  version = "4.0.0";
 
   src = fetchurl {
     url =
       "mirror://apache/arrow/arrow-${version}/apache-arrow-${version}.tar.gz";
-    sha256 = "0yp2b02wrc3s50zd56fmpz4nhhbihp0zw329v4zizaipwlxwrhkk";
+    sha256 = "1bj9jr0pgq9f2nyzqiyj3cl0hcx3c83z2ym6rpdkp59ff2zx0caa";
   };
   sourceRoot = "apache-arrow-${version}/cpp";
 
@@ -90,6 +91,10 @@ in stdenv.mkDerivation rec {
     "-DARROW_VERBOSE_THIRDPARTY_BUILD=ON"
     "-DARROW_DEPENDENCY_SOURCE=SYSTEM"
     "-DARROW_DEPENDENCY_USE_SHARED=${if enableShared then "ON" else "OFF"}"
+    "-DARROW_COMPUTE=ON"
+    "-DARROW_CSV=ON"
+    "-DARROW_DATASET=ON"
+    "-DARROW_JSON=ON"
     "-DARROW_PLASMA=ON"
     # Disable Python for static mode because openblas is currently broken there.
     "-DARROW_PYTHON=${if enableShared then "ON" else "OFF"}"
@@ -111,6 +116,8 @@ in stdenv.mkDerivation rec {
     "-DCMAKE_INSTALL_RPATH=@loader_path/../lib" # needed for tools executables
   ] ++ lib.optional (!stdenv.isx86_64) "-DARROW_USE_SIMD=OFF";
 
+  ARROW_XSIMD_URL = xsimd.src;
+
   doInstallCheck = true;
   ARROW_TEST_DATA =
     if doInstallCheck then "${arrow-testing}/data" else null;

pkgs/development/libraries/xsimd/default.nix

@@ -0,0 +1,56 @@
+{ lib, stdenv, fetchFromGitHub, cmake, gtest }:
+let
+  version = "7.5.0";
+
+  darwin_src = fetchFromGitHub {
+    owner = "xtensor-stack";
+    repo = "xsimd";
+    rev = version;
+    sha256 = "eGAdRSYhf7rbFdm8g1Tz1ZtSVu44yjH/loewblhv9Vs=";
+    # Avoid requiring apple_sdk. We're doing this here instead of in the patchPhase
+    # because this source is directly used in arrow-cpp.
+    # pyconfig.h defines _GNU_SOURCE to 1, so we need to stamp that out too.
+    # Upstream PR with a better fix: https://github.com/xtensor-stack/xsimd/pull/463
+    postFetch = ''
+      mkdir $out
+      tar -xf $downloadedFile --directory=$out --strip-components=1
+      substituteInPlace $out/include/xsimd/types/xsimd_scalar.hpp \
+        --replace 'defined(__APPLE__)' 0 \
+        --replace 'defined(_GNU_SOURCE)' 0
+    '';
+  };
+
+  src = fetchFromGitHub {
+    owner = "xtensor-stack";
+    repo = "xsimd";
+    rev = version;
+    sha256 = "0c9pq5vz43j99z83w3b9qylfi66mn749k1afpv5cwfxggbxvy63f";
+  };
+in stdenv.mkDerivation {
+  pname = "xsimd";
+  inherit version;
+  src = if stdenv.hostPlatform.isDarwin then darwin_src else src;
+
+  nativeBuildInputs = [ cmake ];
+
+  cmakeFlags = [ "-DBUILD_TESTS=ON" ];
+
+  doCheck = true;
+  checkInputs = [ gtest ];
+  checkTarget = "xtest";
+  GTEST_FILTER = let
+      # Upstream Issue: https://github.com/xtensor-stack/xsimd/issues/456
+      filteredTests = lib.optionals stdenv.hostPlatform.isDarwin [
+        "error_gamma_test/sse_double.gamma"
+        "error_gamma_test/avx_double.gamma"
+      ];
+    in "-${builtins.concatStringsSep ":" filteredTests}";
+
+  meta = with lib; {
+    description = "C++ wrappers for SIMD intrinsics";
+    homepage = "https://github.com/xtensor-stack/xsimd";
+    license = licenses.bsd3;
+    maintainers = with maintainers; [ tobim ];
+    platforms = platforms.all;
+  };
+}

pkgs/development/python-modules/pyarrow/default.nix

@@ -34,12 +34,17 @@ buildPythonPackage rec {
     export PYARROW_PARALLEL=$NIX_BUILD_CORES
   '';
 
-  # Deselect a single test because pyarrow prints a 2-line error message where
-  # only a single line is expected. The additional line of output comes from
-  # the glog library which is an optional dependency of arrow-cpp that is
-  # enabled in nixpkgs.
-  # Upstream Issue: https://issues.apache.org/jira/browse/ARROW-11393
-  pytestFlagsArray = [ "--deselect=pyarrow/tests/test_memory.py::test_env_var" ];
+  pytestFlagsArray = [
+    # Deselect a single test because pyarrow prints a 2-line error message where
+    # only a single line is expected. The additional line of output comes from
+    # the glog library which is an optional dependency of arrow-cpp that is
+    # enabled in nixpkgs.
+    # Upstream Issue: https://issues.apache.org/jira/browse/ARROW-11393
+    "--deselect=pyarrow/tests/test_memory.py::test_env_var"
+    # Deselect the parquet dataset write test because it erroneously fails to find the
+    # pyarrow._dataset module.
+    "--deselect=pyarrow/tests/parquet/test_dataset.py::test_write_to_dataset_filesystem"
+  ];
 
   dontUseSetuptoolsCheck = true;
   preCheck = ''

pkgs/top-level/all-packages.nix

@@ -18104,6 +18104,8 @@ in
 
   xlslib = callPackage ../development/libraries/xlslib { };
 
+  xsimd = callPackage ../development/libraries/xsimd { };
+
   xvidcore = callPackage ../development/libraries/xvidcore { };
 
   xxHash = callPackage ../development/libraries/xxHash {};