You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@arrow.apache.org by we...@apache.org on 2018/09/14 18:56:29 UTC

[arrow] branch master updated: ARROW-3229: [Packaging]: Adjust wheel package scripts to account for Parquet codebase migration

This is an automated email from the ASF dual-hosted git repository.

wesm pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/arrow.git


The following commit(s) were added to refs/heads/master by this push:
     new c48eaec  ARROW-3229: [Packaging]: Adjust wheel package scripts to account for Parquet codebase migration
c48eaec is described below

commit c48eaec92a4af62c5e6f02059b6ce787cffb9fc0
Author: Krisztián Szűcs <sz...@gmail.com>
AuthorDate: Fri Sep 14 14:56:22 2018 -0400

    ARROW-3229: [Packaging]: Adjust wheel package scripts to account for Parquet codebase migration
    
    See the build results here: https://github.com/kszucs/crossbow/branches/all?utf8=%E2%9C%93&query=build-308
    
    Manylinux1 wheels are already [updated](https://github.com/apache/arrow/commit/9b4cd9c03ed9365f8e235f296caa166ea692c98f#diff-785c937d41d0bf0c426191906ee42039)
    
    Author: Krisztián Szűcs <sz...@gmail.com>
    
    Closes #2552 from kszucs/ARROW-3229 and squashes the following commits:
    
    6323fe7fa <Krisztián Szűcs> disable both arrow and 3rdparty verbose builds
    7d5a0ede9 <Krisztián Szűcs> don't build thrift manually
    f7e533131 <Krisztián Szűcs> adjust wheel builds
---
 dev/tasks/python-wheels/appveyor.yml  |  1 -
 dev/tasks/python-wheels/osx-build.sh  | 50 ++---------------------------------
 dev/tasks/python-wheels/win-build.bat | 19 +------------
 python/manylinux1/build_arrow.sh      |  7 ++++-
 4 files changed, 9 insertions(+), 68 deletions(-)

diff --git a/dev/tasks/python-wheels/appveyor.yml b/dev/tasks/python-wheels/appveyor.yml
index d9d208c..016041a 100644
--- a/dev/tasks/python-wheels/appveyor.yml
+++ b/dev/tasks/python-wheels/appveyor.yml
@@ -26,7 +26,6 @@ environment:
   ARROW_SRC: C:\apache-arrow
   PYARROW_VERSION: {{ arrow.version }}
   PYARROW_REF: {{ arrow.head }}
-  PARQUET_CPP_REF: master
 
 init:
   - set MINICONDA=C:\Miniconda35-x64
diff --git a/dev/tasks/python-wheels/osx-build.sh b/dev/tasks/python-wheels/osx-build.sh
index 6dce16f..4170a93 100755
--- a/dev/tasks/python-wheels/osx-build.sh
+++ b/dev/tasks/python-wheels/osx-build.sh
@@ -93,37 +93,6 @@ function build_wheel {
       install_name_tool -change libarrow_boost_system.dylib @rpath/libarrow_boost_system.dylib libarrow_boost_filesystem.dylib
     popd
 
-    # We build a custom version of thrift instead of using the one that comes
-    # with brew as we also want it to use our namespaced version of Boost.
-    # TODO(PARQUET-1262): Use the external project facilities of parquet-cpp.
-    export THRIFT_HOME=`pwd`/thift-dist
-    export THRIFT_VERSION=0.11.0
-    wget http://archive.apache.org/dist/thrift/${THRIFT_VERSION}/thrift-${THRIFT_VERSION}.tar.gz
-    tar xf thrift-${THRIFT_VERSION}.tar.gz
-    pushd thrift-${THRIFT_VERSION}
-    mkdir build-tmp
-    pushd build-tmp
-    cmake -DCMAKE_BUILD_TYPE=release \
-        "-DCMAKE_CXX_FLAGS=-fPIC" \
-        "-DCMAKE_C_FLAGS=-fPIC" \
-        "-DCMAKE_INSTALL_PREFIX=${THRIFT_HOME}" \
-        "-DCMAKE_INSTALL_RPATH=${THRIFT_HOME}/lib" \
-        "-DBUILD_SHARED_LIBS=OFF" \
-        "-DBUILD_TESTING=OFF" \
-        "-DWITH_QT4=OFF" \
-        "-DWITH_C_GLIB=OFF" \
-        "-DWITH_JAVA=OFF" \
-        "-DWITH_PYTHON=OFF" \
-        "-DWITH_CPP=ON" \
-        "-DWITH_STATIC_LIB=ON" \
-        "-DWITH_LIBEVENT=OFF" \
-        -DBoost_NAMESPACE=arrow_boost \
-        -DBOOST_ROOT="$arrow_boost_dist" \
-        ..
-    make install -j5
-    popd
-    popd
-
     # Now we can start with the actual build of Arrow and Parquet.
     # We pin NumPy to an old version here as the NumPy version one builds
     # with is the oldest supported one. Thanks to NumPy's guarantees our Arrow
@@ -131,6 +100,7 @@ function build_wheel {
     export ARROW_HOME=`pwd`/arrow-dist
     export PARQUET_HOME=`pwd`/arrow-dist
     pip install "cython==0.27.3" "numpy==${NP_TEST_DEP}"
+
     pushd cpp
     mkdir build
     pushd build
@@ -143,6 +113,7 @@ function build_wheel {
           -DARROW_PLASMA=ON \
           -DARROW_RPATH_ORIGIN=ON \
           -DARROW_PYTHON=ON \
+          -DARROW_PARQUET=ON \
           -DARROW_ORC=ON \
           -DBOOST_ROOT="$arrow_boost_dist" \
           -DBoost_NAMESPACE=arrow_boost \
@@ -153,23 +124,6 @@ function build_wheel {
     popd
     popd
 
-    git clone https://github.com/apache/parquet-cpp.git
-    pushd parquet-cpp
-    mkdir build
-    pushd build
-    cmake -DCMAKE_BUILD_TYPE=Release \
-          -DCMAKE_INSTALL_PREFIX=$PARQUET_HOME \
-          -DPARQUET_VERBOSE_THIRDPARTY_BUILD=ON \
-          -DPARQUET_BUILD_TESTS=OFF \
-          -DPARQUET_BOOST_USE_SHARED=ON \
-          -DBoost_NAMESPACE=arrow_boost \
-          -DBOOST_ROOT="$arrow_boost_dist" \
-          ..
-    make -j5 VERBOSE=1
-    make install
-    popd
-    popd
-
     # Unset the HOME variables and use pkg-config to discover the previously
     # built binaries. By using pkg-config, we also are able to discover the
     # ABI and SO versions of the dynamic libraries.
diff --git a/dev/tasks/python-wheels/win-build.bat b/dev/tasks/python-wheels/win-build.bat
index 37d0b64..c73b39b 100644
--- a/dev/tasks/python-wheels/win-build.bat
+++ b/dev/tasks/python-wheels/win-build.bat
@@ -42,7 +42,6 @@ popd
 set ARROW_HOME=%CONDA_PREFIX%\Library
 set PARQUET_HOME=%CONDA_PREFIX%\Library
 set ARROW_BUILD_TOOLCHAIN=%CONDA_PREFIX%\Library
-set PARQUET_BUILD_TOOLCHAIN=%CONDA_PREFIX%\Library
 
 echo %ARROW_HOME%
 
@@ -57,6 +56,7 @@ cmake -G "%GENERATOR%" ^
       -DCMAKE_BUILD_TYPE=Release ^
       -DARROW_CXXFLAGS="/MP" ^
       -DARROW_PYTHON=ON ^
+      -DARROW_PARQUET=ON ^
       ..  || exit /B
 cmake --build . --target INSTALL --config Release  || exit /B
 
@@ -65,23 +65,6 @@ set PYTHONPATH=%CONDA_PREFIX%\Lib;%CONDA_PREFIX%\Lib\site-packages;%CONDA_PREFIX
 ctest -VV  || exit /B
 popd
 
-@rem Build parquet-cpp
-git clone https://github.com/apache/parquet-cpp.git || exit /B
-pushd parquet-cpp
-git checkout "%PARQUET_CPP_REF%"
-popd
-
-mkdir parquet-cpp\build
-pushd parquet-cpp\build
-
-cmake -G "%GENERATOR%" ^
-     -DCMAKE_INSTALL_PREFIX=%PARQUET_HOME% ^
-     -DCMAKE_BUILD_TYPE=Release ^
-     -DPARQUET_BOOST_USE_SHARED=OFF ^
-     -DPARQUET_BUILD_TESTS=OFF .. || exit /B
-cmake --build . --target INSTALL --config Release || exit /B
-popd
-
 @rem Build and import pyarrow
 set PYTHONPATH=
 
diff --git a/python/manylinux1/build_arrow.sh b/python/manylinux1/build_arrow.sh
index 1313842..04034e7 100755
--- a/python/manylinux1/build_arrow.sh
+++ b/python/manylinux1/build_arrow.sh
@@ -90,7 +90,12 @@ for PYTHON_TUPLE in ${PYTHON_VERSIONS}; do
     # Clear output directory
     rm -rf dist/
     echo "=== (${PYTHON}) Building wheel ==="
-    PATH="$PATH:${CPYTHON_PATH}/bin" $PYTHON_INTERPRETER setup.py build_ext --inplace --with-parquet --bundle-arrow-cpp --bundle-boost --boost-namespace=arrow_boost
+    PATH="$PATH:${CPYTHON_PATH}/bin" $PYTHON_INTERPRETER setup.py build_ext \
+        --inplace \
+        --with-parquet \
+        --bundle-arrow-cpp \
+        --bundle-boost \
+        --boost-namespace=arrow_boost
     PATH="$PATH:${CPYTHON_PATH}/bin" $PYTHON_INTERPRETER setup.py bdist_wheel
     PATH="$PATH:${CPYTHON_PATH}/bin" $PYTHON_INTERPRETER setup.py sdist