Copy tracker

.github/workflows/linux.yml

@@ -143,6 +143,7 @@ jobs:
           -DCMAKE_CXX_STANDARD=${{matrix.cpp_standard}} \
           -DUSE_DATE_POLYFILL=${{matrix.date-polyfill}} \
           -DBUILD_TESTS=ON \
+          -DTRACK_COPIES=ON \
           -DENABLE_INTEGRATION_TEST=${{matrix.shared}} \
           -DBUILD_EXAMPLES=ON \
           -DCREATE_JSON_READER_TARGET=ON \

.github/workflows/osx.yml

@@ -68,6 +68,7 @@ jobs:
           -DCMAKE_BUILD_TYPE:STRING=${{matrix.config.name}} \
           -DCMAKE_INSTALL_PREFIX=$CONDA_PREFIX \
           -DBUILD_TESTS=ON \
+          -DTRACK_COPIES=ON \
           -DENABLE_INTEGRATION_TEST=ON \
           -DBUILD_EXAMPLES=ON \
           -DCMAKE_C_COMPILER_LAUNCHER=ccache \
@@ -112,4 +113,4 @@ jobs:
 
       - name: Install
         working-directory: build
-        run: cmake --install . --config ${{matrix.config.name}}
+        run: cmake --install . --config ${{matrix.config.name}}

.github/workflows/windows.yml

@@ -134,6 +134,7 @@ jobs:
         -DCMAKE_PREFIX_PATH=$SPARROW_DEPS_PREFIX \
         -DCMAKE_INSTALL_PREFIX=$SPARROW_INSTALL_PREFIX \
         -DBUILD_TESTS=ON \
+        -DTRACK_COPIES=ON \
         -DENABLE_INTEGRATION_TEST=${{matrix.toolchain.shared}} \
         -DBUILD_EXAMPLES=ON \
         -DCREATE_JSON_READER_TARGET=ON \

CMakeLists.txt

@@ -103,6 +103,7 @@ MESSAGE(STATUS "🔧 Enable integration tests: ${ENABLE_INTEGRATION_TEST}")
 OPTION(CREATE_JSON_READER_TARGET "Create json_reader target, usualy used for integration tests" OFF)
 OPTION(ENABLE_INTEGRATION_TEST "Creates json_reader target and enable integration tests" OFF)
 OPTION(CREATE_JSON_READER_TARGET "Create json_reader target, automatically set when ENABLE_INTEGRATION_TEST is ON" OFF)
+OPTION(TRACK_COPIES, "Track copies in tests" OFF)
 
 if(ENABLE_INTEGRATION_TEST)
     set(CREATE_JSON_READER_TARGET ON)
@@ -182,6 +183,11 @@ if(SPARROW_CONTRACTS_THROW_ON_FAILURE)
     list(APPEND SPARROW_COMPILE_DEFINITIONS SPARROW_CONTRACTS_THROW_ON_FAILURE=1)
 endif()
 
+if(TRACK_COPIES)
+    message(STATUS "Tracking copies in tests")
+    list(APPEND SPARROW_COMPILE_DEFINITIONS SPARROW_TRACK_COPIES)
+endif()
+
 # Build
 # =====
 set(BINARY_BUILD_DIR "${CMAKE_BINARY_DIR}/bin/${CMAKE_BUILD_TYPE}")
@@ -228,6 +234,9 @@ set(SPARROW_HEADERS
     ${SPARROW_INCLUDE_DIR}/sparrow/config/config.hpp
     ${SPARROW_INCLUDE_DIR}/sparrow/config/sparrow_version.hpp
 
+    # debug
+    ${SPARROW_INCLUDE_DIR}/sparrow/debug/copy_tracker.hpp
+
     # detail
     ${SPARROW_INCLUDE_DIR}/sparrow/details/3rdparty/float16_t.hpp
 
@@ -324,6 +333,7 @@ set(SPARROW_SRC
     ${SPARROW_SOURCE_DIR}/arrow_interface/arrow_array_schema_proxy.cpp
     ${SPARROW_SOURCE_DIR}/arrow_interface/arrow_schema.cpp
     ${SPARROW_SOURCE_DIR}/arrow_interface/private_data_ownership.cpp
+    ${SPARROW_SOURCE_DIR}/debug/copy_tracker.cpp
     ${SPARROW_SOURCE_DIR}/layout/array_factory.cpp
     ${SPARROW_SOURCE_DIR}/layout/array_helper.cpp
     ${SPARROW_SOURCE_DIR}/layout/fixed_width_binary_array_utils.cpp

include/sparrow/debug/copy_tracker.hpp

@@ -0,0 +1,28 @@
+#pragma once
+
+#include <string>
+#include <vector>
+
+#include "sparrow/config/config.hpp"
+
+namespace sparrow::copy_tracker
+{
+    template <class T>
+    std::string key();
+
+    constexpr bool is_enabled()
+    {
+#ifdef SPARROW_TRACK_COPIES
+        return true;
+#else
+        return false;
+#endif
+    }
+
+    SPARROW_API void increase(const std::string& key);
+    SPARROW_API void reset(const std::string& key);
+    SPARROW_API void reset_all();
+    SPARROW_API int count(const std::string& key, int disabled_value = 0);
+    SPARROW_API std::vector<std::string> key_list();
+
+}

src/debug/copy_tracker.cpp

@@ -0,0 +1,105 @@
+#include "sparrow/debug/copy_tracker.hpp"
+
+#include <map>
+#include <ranges>
+#include <string>
+
+namespace sparrow::copy_tracker
+{
+#ifdef SPARROW_TRACK_COPIES
+
+    class copy_tracker_impl
+    {
+    public:
+
+        static copy_tracker_impl& instance()
+        {
+            static copy_tracker_impl inst;
+            return inst;
+        }
+
+        void increase(const std::string& key)
+        {
+            m_copy_count[key] += 1;
+        }
+
+        void reset(const std::string& key)
+        {
+            m_copy_count[key] = 0;
+        }
+
+        void reset_all()
+        {
+            m_copy_count.clear();
+        }
+
+        int count(const std::string& key)
+        {
+            return m_copy_count[key];
+        }
+
+        auto key_list()
+        {
+            return std::views::keys(m_copy_count);
+        }
+
+    private:
+
+        copy_tracker_impl() = default;
+
+        std::map<std::string, int> m_copy_count = {};
+    };
+
+    void increase(const std::string& key)
+    {
+        return copy_tracker_impl::instance().increase(key);
+    }
+
+    void reset(const std::string& key)
+    {
+        return copy_tracker_impl::instance().reset(key);
+    }
+
+    void reset_all()
+    {
+        return copy_tracker_impl::instance().reset_all();
+    }
+
+    int count(const std::string& key, int /*disabled_value*/)
+    {
+        return copy_tracker_impl::instance().count(key);
+    }
+
+    std::vector<std::string> key_list()
+    {
+        auto keys = copy_tracker_impl::instance().key_list();
+        return {keys.begin(), keys.end()};
+    }
+
+#else
+
+    void increase(const std::string&)
+    {
+    }
+
+    void reset(const std::string&)
+    {
+    }
+
+    void reset_all()
+    {
+    }
+
+    int count(const std::string&, int disabled_value)
+    {
+        return disabled_value;
+    }
+
+    std::vector<std::string> key_list()
+    {
+        return {};
+    }
+
+#endif
+
+}

src/record_batch.cpp

@@ -16,10 +16,20 @@
 
 #include <unordered_set>
 
+#include "sparrow/debug/copy_tracker.hpp"
 #include "sparrow/utils/contracts.hpp"
 
 namespace sparrow
 {
+    namespace copy_tracker
+    {
+        template <>
+        SPARROW_API std::string key<record_batch>()
+        {
+            return "record_batch";
+        }
+    }
+
     record_batch::record_batch(initializer_type init)
     {
         m_name_list.reserve(init.size());
@@ -97,6 +107,7 @@ namespace sparrow
         , m_array_list(rhs.m_array_list)
     {
         update_array_map_cache();
+        copy_tracker::increase(copy_tracker::key<record_batch>());
     }
 
     record_batch& record_batch::operator=(const record_batch& rhs)

test/test_record_batch.cpp

@@ -14,6 +14,7 @@
 
 #include <string_view>
 
+#include "sparrow/debug/copy_tracker.hpp"
 #include "sparrow/primitive_array.hpp"
 #include "sparrow/record_batch.hpp"
 
@@ -199,9 +200,16 @@ namespace sparrow
 
         TEST_CASE("copy semantic")
         {
+#ifdef SPARROW_TRACK_COPIES
+            auto key = std::string("record_batch");  // copy_tracker::key<record_batch>();
+            copy_tracker::reset(key);
+#endif
             auto record1 = make_record_batch(col_size);
             auto record2(record1);
             CHECK_EQ(record1, record2);
+#ifdef SPARROW_TRACK_COPIES
+            CHECK_EQ(copy_tracker::count(key), 1);
+#endif
 
             auto record3 = make_record_batch(col_size + 2u);
             CHECK_NE(record1, record3);
@@ -330,6 +338,17 @@ namespace sparrow
             );
             CHECK(res);
         }
+#ifdef SPARROW_TRACK_COPIES
+        TEST_CASE("emplace_back")
+        {
+            std::vector<record_batch> vec(3, make_record_batch(col_size));
+            std::cout << "capacity: " << vec.capacity() << std::endl;
+            auto key = std::string("record_batch");  // copy_tracker::key<record_batch>();
+            copy_tracker::reset(key);
+            vec.emplace_back(make_record_batch(col_size));
+            CHECK_EQ(copy_tracker::count(key), 0);
+        }
+#endif
 
 #if defined(__cpp_lib_format)
         TEST_CASE("formatter")