Investigate slowdown in 4.5.8

.github/workflows/conda.yml

@@ -17,55 +17,55 @@ defaults:
     shell: bash -el {0}
 
 jobs:
-  macos-tests:
-    runs-on: macos-latest
-    strategy:
-      fail-fast: false
-      matrix:
-        PYTHON_VERSION: ['3.9', '3.10', '3.11']
-        PYARROW_VERSION: ['10.0.1']
-        include:
-          - PYTHON_VERSION: "3.10"
-            PYARROW_VERSION: "9.0.0"
-          - PYTHON_VERSION: "3.11"
-            PYARROW_VERSION: "12.0.0"
-          - PYTHON_VERSION: "3.11"
-            PYARROW_VERSION: "13.0.0"
-    steps:
-      - name: Checkout branch
-        uses: actions/checkout@v3
-        with:
-          submodules: true
-      - name: Fetch full git history
-        run: git fetch --prune --unshallow
-      - name: Set up Conda env
-        uses: mamba-org/provision-with-micromamba@e2b397b12d0a38069451664382b769c9456e3d6d
-        with:
-          condarc-file: .github/workflows/.condarc
-          cache-env: true
-          extra-specs: |
-            python=${{ matrix.PYTHON_VERSION }}
-            pyarrow=${{ matrix.PYARROW_VERSION }}
-            pytest-md
-            pytest-emoji
-      - name: Build with CMake
-        run: |
-          mkdir build && \
-          cd build && \
-          cmake -DBOOST_ROOT=${CONDA_PREFIX} -DBUILD_COVERAGE=ON \
-            -DCMAKE_INSTALL_PREFIX=./dist  \
-            -DPYTHON_EXECUTABLE=$(which python) \
-            -GNinja .. && \
-          ninja && \
-          cmake --build . --target install
-      - name: Install repository
-        run: |
-          cd build/dist && python -m pip install --no-build-isolation --no-deps --disable-pip-version-check -e .
-      - name: Test import
-        run: |
-          cd /
-          python -c 'import turbodbc'
-          python -c "import turbodbc.arrow_support"
+#  macos-tests:
+#    runs-on: macos-latest
+#    strategy:
+#      fail-fast: false
+#      matrix:
+#        PYTHON_VERSION: ['3.9', '3.10', '3.11']
+#        PYARROW_VERSION: ['10.0.1']
+#        include:
+#          - PYTHON_VERSION: "3.10"
+#            PYARROW_VERSION: "9.0.0"
+#          - PYTHON_VERSION: "3.11"
+#            PYARROW_VERSION: "12.0.0"
+#          - PYTHON_VERSION: "3.11"
+#            PYARROW_VERSION: "13.0.0"
+#    steps:
+#      - name: Checkout branch
+#        uses: actions/checkout@v3
+#        with:
+#          submodules: true
+#      - name: Fetch full git history
+#        run: git fetch --prune --unshallow
+#      - name: Set up Conda env
+#        uses: mamba-org/provision-with-micromamba@e2b397b12d0a38069451664382b769c9456e3d6d
+#        with:
+#          condarc-file: .github/workflows/.condarc
+#          cache-env: true
+#          extra-specs: |
+#            python=${{ matrix.PYTHON_VERSION }}
+#            pyarrow=${{ matrix.PYARROW_VERSION }}
+#            pytest-md
+#            pytest-emoji
+#      - name: Build with CMake
+#        run: |
+#          mkdir build && \
+#          cd build && \
+#          cmake -DBOOST_ROOT=${CONDA_PREFIX} -DBUILD_COVERAGE=ON \
+#            -DCMAKE_INSTALL_PREFIX=./dist  \
+#            -DPYTHON_EXECUTABLE=$(which python) \
+#            -GNinja .. && \
+#          ninja && \
+#          cmake --build . --target install
+#      - name: Install repository
+#        run: |
+#          cd build/dist && python -m pip install --no-build-isolation --no-deps --disable-pip-version-check -e .
+#      - name: Test import
+#        run: |
+#          cd /
+#          python -c 'import turbodbc'
+#          python -c "import turbodbc.arrow_support"
   windows-tests:
     runs-on: windows-latest
     defaults:
@@ -127,8 +127,11 @@ jobs:
           dir src
           dir src\turbodbc_arrow
           if %errorlevel% neq 0 exit /b %errorlevel%
-          python -m pip install --no-build-isolation --no-deps --disable-pip-version-check -e .
+          set "BOOST_ROOT=%CONDA_PREFIX%\Library"
+          python -m pip install --no-build-isolation --no-deps --disable-pip-version-check .
           if %errorlevel% neq 0 exit /b %errorlevel%
+          @CALL micromamba install build
+          python -m build --outdir ./dist
       - name: Test import
         run: |
           @CALL micromamba activate turbodbc

cmake_scripts/default_build_setup.cmake

@@ -64,3 +64,11 @@ add_custom_target(refresh_cmake_configuration
 	cmake -E touch ${CMAKE_PARENT_LIST_FILE} # make cmake detect configuration is changed on NEXT build
 	COMMENT "Forcing refreshing of the CMake configuration. This allows to use globbing safely."
 )
+
+if(WIN32)
+    add_definitions("-std=c++17")
+    link_directories("$ENV{PYTHON}/libs")
+    set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS "TRUE")
+    # set(Boost_USE_STATIC_RUNTIME "ON")
+    # set(Boost_USE_STATIC_LIBS "ON")
+endif()

cpp/cpp_odbc/CMakeLists.txt

@@ -1,3 +1,6 @@
+find_package(Boost REQUIRED COMPONENTS locale)
+include_directories(SYSTEM ${Boost_INCLUDE_DIRS})
+
 find_package(odbc REQUIRED)
 include_directories(SYSTEM ${Odbc_INCLUDE_DIR})
 

cpp/cpp_odbc/Library/src/level2/level1_connector.cpp

@@ -20,6 +20,8 @@
 #include <locale>
 #include <sstream>
 
+#include <boost/locale.hpp>
+
 namespace impl {
 
 template <typename Output_Handle, typename Input_Handle>
@@ -430,7 +432,12 @@ column_description level1_connector::do_describe_column_wide(statement_handle co
 
     impl::throw_on_error(return_code, *this, handle);
 
-    std::string const utf8_name = std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t>{}.to_bytes(name);
+    // TODO-Simeon: throws
+    // throw std::invalid_argument("Simeon conversion in level1_connector");
+    // TODO-Simeon: throws
+
+    //std::string const utf8_name = std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t>{}.to_bytes(name);
+    auto const utf8_name = boost::locale::conv::utf_to_utf<char>(static_cast<std::u16string>(name));
     return {utf8_name, data_type, size, decimal_digits, allows_nullable};
 }
 

cpp/turbodbc/CMakeLists.txt

@@ -1,4 +1,4 @@
-find_package(Boost REQUIRED COMPONENTS system date_time)
+find_package(Boost REQUIRED COMPONENTS system date_time locale)
 include_directories(SYSTEM ${Boost_INCLUDE_DIRS})
 
 find_package(odbc REQUIRED)

cpp/turbodbc/Library/src/connect.cpp

@@ -1,6 +1,8 @@
 #include "turbodbc/connect.h"
 #include <cpp_odbc/make_environment.h>
 
+using namespace std;
+
 turbodbc::connection turbodbc::connect(std::string const & connection_string, options options)
 {
 	auto environment = cpp_odbc::make_environment();

cpp/turbodbc/Library/src/cursor.cpp

@@ -14,12 +14,19 @@
 #include <sstream>
 #include <codecvt>
 
+#include <boost/locale.hpp>
+
 namespace {
 
     std::u16string as_utf16(std::string utf8_encoded) {
+        // TODO-Simeon: throws
+        // throw std::invalid_argument("Simeon conversion in cursor");
+        // TODO-Simeon: throws
+
         // not all compilers support the new C++11 conversion facets
-        static std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> converter;
-        return converter.from_bytes(utf8_encoded);
+        return boost::locale::conv::utf_to_utf<char16_t>(utf8_encoded);
+        //static std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> converter;
+        //return converter.from_bytes(utf8_encoded);
     }
 
 }

cpp/turbodbc_arrow/Library/src/arrow_result_set.cpp

@@ -15,6 +15,8 @@
 #include <codecvt>
 #include <vector>
 
+#include <boost/locale.hpp>
+
 using arrow::default_memory_pool;
 using arrow::AdaptiveIntBuilder;
 using arrow::ArrayBuilder;
@@ -127,7 +129,11 @@ Status AppendUnicodeStringsToBuilder(size_t rows_in_batch, BuilderType& builder,
         if (element.indicator == SQL_NULL_DATA) {
             ARROW_RETURN_NOT_OK(builder.AppendNullProxy());
         } else {
+            // TODO-Simeon: throws
+            // throw std::invalid_argument("Simeon conversion in arrow_result_set");
+            // TODO-Simeon: throws
             std::u16string str_u16(reinterpret_cast<const char16_t*>(element.data_pointer), element.indicator / 2);
+            //std::string u8string = std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t>{}.to_bytes(str_u16);
             std::string u8string = std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t>{}.to_bytes(str_u16);
             ARROW_RETURN_NOT_OK(builder.AppendProxy(u8string.data(), u8string.size()));
         }

cpp/turbodbc_arrow/Library/src/set_arrow_parameters.cpp

@@ -14,6 +14,8 @@
 #include <ciso646>
 #include <codecvt>
 
+#include <boost/locale.hpp>
+
 using arrow::BooleanArray;
 using arrow::BinaryArray;
 using arrow::ChunkedArray;

environment.yml

@@ -4,7 +4,7 @@ channels:
 dependencies:
   - python>=3.9
   - pyarrow>=7.0,<14
-  - sel(unix): unixodbc
+  # - sel(unix): unixodbc
   - c-compiler
   - cxx-compiler
   - gmock=1.10.0

setup.py

@@ -104,7 +104,12 @@ def __str__(self):
     extra_compile_args.append("/std:c++17")
     boost_root = os.getenv("BOOST_ROOT")
     if boost_root:
+        # raise Exception("Hey")
+        library_dirs.append(boost_root)
+        library_dirs.append(os.path.join(boost_root, "lib"))
+
         include_dirs.append(boost_root)
+        library_dirs.append(boost_root)
         library_dirs.append(os.path.join(boost_root, "stage", "lib"))
         library_dirs.append(os.path.join(boost_root, "lib64-msvc-14.0"))
     else: