Setup workflows and tests

Fix any bugs that fail initial cross-platform tests

.github/dependabot.yml

@@ -0,0 +1,8 @@
+version: 2
+updates:
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      # Check for updates to GitHub Actions every week
+      # See https://docs.github.com/en/code-security/dependabot/working-with-dependabot/keeping-your-actions-up-to-date-with-dependabot
+      interval: "weekly"

.github/workflows/tests.yml

@@ -0,0 +1,93 @@
+name: tests
+
+on:
+  push:
+  pull_request:
+
+jobs:
+  build:
+    name: ${{matrix.os}} ${{ matrix.description }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - os: ubuntu-20.04
+            # oldest LLVM that the install-llvm-action action is able to install
+            description: "LLVM 5"
+            llvm-version: "5"
+
+          - os: ubuntu-20.04
+            # oldest GCC that the setup-gcc action is able to install
+            description: "GCC 7"
+            gcc-version: "7"
+            cmake-version: "3.18"
+
+          - os: ubuntu-latest
+            description: "GCC 13"
+            gcc-version: "13"
+
+          - os: ubuntu-latest
+            # default GCC, which has gcov
+
+          - os: macos-latest
+            # uses Apple Clang
+
+          - os: windows-latest
+            # uses MSVC
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install TBB (Ubuntu)
+        if: contains(matrix.os, 'ubuntu')
+        run: sudo apt-get install -y libtbb-dev
+
+      - name: Install TBB (macOS)
+        if: contains(matrix.os, 'macos')
+        run: brew install tbb
+
+      - name: Setup GCC
+        uses: egor-tensin/setup-gcc@v1
+        if: ${{ matrix.gcc-version != '' }}
+        with:
+          version: ${{ matrix.gcc-version }}
+
+      - name: Setup LLVM and Clang
+        uses: KyleMayes/install-llvm-action@v1
+        if: ${{ matrix.llvm-version != '' }}
+        with:
+          version: ${{ matrix.llvm-version }}
+          env: true
+
+      - name: Setup cmake
+        uses: jwlawson/actions-setup-cmake@v1
+        if: ${{ matrix.cmake-version != '' }}
+        with:
+          cmake-version: ${{ matrix.cmake-version }}
+
+      - name: Build and Test
+        run: |
+          cmake -S . -B build/ -DCMAKE_BUILD_TYPE=Debug -DPOOLSTL_TEST=ON -DPOOLSTL_TEST_COVERAGE=ON
+          cmake --build build/ --config Debug
+          cd build/tests
+          ctest -C Debug --output-on-failure --verbose
+          cd ../..
+          find .
+        shell: bash
+
+      - name: Benchmark
+        if: ${{ matrix.gcc-version != '7' }}
+        run: |
+          cmake -S . -B bench_build/ -DCMAKE_BUILD_TYPE=Release -DPOOLSTL_BENCH=ON
+          cmake --build bench_build/ --config Release
+          cd bench_build/benchmark/
+          ./poolstl_bench || ./Release/poolstl_bench.exe
+        shell: bash
+
+      - name: Upload Coverage to Codecov
+        if: contains(matrix.os, 'ubuntu')
+        uses: codecov/codecov-action@v3
+        with:
+          gcov: true
+          gcov_include: include/*

README.md

@@ -1,6 +1,9 @@
+[![tests](https://github.com/alugowski/poolSTL/actions/workflows/tests.yml/badge.svg)](https://github.com/alugowski/poolSTL/actions/workflows/tests.yml)
+[![codecov](https://codecov.io/gh/alugowski/poolSTL/branch/main/graph/badge.svg?token=)](https://codecov.io/gh/alugowski/poolSTL)
+
 # poolSTL
 
-Thread pool-based implementation of [parallel standard library algorithms](https://en.cppreference.com/w/cpp/algorithm#Execution_policies).
+Thread pool-based implementation of [parallel standard library algorithms](https://en.cppreference.com/w/cpp/algorithm).
 
 Those algorithms are great, but compiler support is inconsistent.
 PoolSTL is a *supplement* to fill in the support gaps so we can use parallel algorithms now.

benchmark/algorithm_bench.cpp

@@ -45,9 +45,9 @@ void for_each(benchmark::State& state) {
 
     for ([[maybe_unused]] auto _ : state) {
         if constexpr (is_policy<ExecutionPolicy>::value) {
-            std::for_each(policy<ExecutionPolicy>::get(), values.begin(), values.end(), [&](auto v) {dest[v] = v;});
+            std::for_each(policy<ExecutionPolicy>::get(), values.begin(), values.end(), [&](auto v) { slow(); dest[v] = v; });
         } else {
-            std::for_each(values.begin(), values.end(), [&](auto v) {dest[v] = v;});
+            std::for_each(values.begin(), values.end(), [&](auto v) {slow(); dest[v] = v;});
         }
         benchmark::DoNotOptimize(dest);
         benchmark::ClobberMemory();
@@ -69,9 +69,9 @@ void transform(benchmark::State& state) {
 
     for ([[maybe_unused]] auto _ : state) {
         if constexpr (is_policy<ExecutionPolicy>::value) {
-            std::transform(policy<ExecutionPolicy>::get(), values.begin(), values.end(), dest.begin(), [&](auto v) { return v; });
+            std::transform(policy<ExecutionPolicy>::get(), values.begin(), values.end(), dest.begin(), [&](auto v) { slow(); return v; });
         } else {
-            std::transform(values.begin(), values.end(), dest.begin(), [&](auto v) { return v; });
+            std::transform(values.begin(), values.end(), dest.begin(), [&](auto v) { slow(); return v; });
         }
         benchmark::DoNotOptimize(dest);
         benchmark::ClobberMemory();

benchmark/main.cpp

@@ -14,4 +14,14 @@ bool static_init() {
 }
 static bool initialized = static_init();
 
+void slow() {
+    int sum = 0;
+
+    for (int i = 0; i < 10; ++i) {
+        sum += i*31 + sum;
+    }
+
+    benchmark::DoNotOptimize(sum);
+}
+
 BENCHMARK_MAIN();

benchmark/utils.hpp

@@ -8,7 +8,9 @@
 #define POOLSTL_BENCH_UTILS_HPP
 
 #include <algorithm>
+#ifdef POOLSTL_BENCH_STD_PAR
 #include <execution>
+#endif
 #include <numeric>
 #include <utility>
 #include <vector>
@@ -40,11 +42,13 @@ template <> struct policy<poolstl::execution::par_pool> {
         return poolstl::execution::par_pool(pool);
     };
 };
+#ifdef POOLSTL_BENCH_STD_PAR
 template <> struct policy<std_par> {
     constexpr static const std::execution::parallel_policy& get() {
         return std::execution::par;
     };
 };
+#endif
 
 template <typename T=int>
 std::vector<T> iota_vector(size_t size, T init=0) {
@@ -53,4 +57,6 @@ std::vector<T> iota_vector(size_t size, T init=0) {
     return ret;
 }
 
+void slow();
+
 #endif

include/poolstl/algorithm

@@ -24,11 +24,11 @@ namespace std {
         auto chunk_size = get_chunk_size(first, last, pool(policy).get_num_threads());
 
         while (first < last) {
-            InputIt loop_end = ::std::min(first + chunk_size, last);
+            InputIt loop_end = first + ::std::min(chunk_size, std::distance(first, last));
 
-            futures.emplace_back(pool(policy).submit([&](InputIt chunk_start, InputIt chunk_stop) {
-                for (; chunk_start != chunk_stop; ++chunk_start) {
-                    f(*chunk_start);
+            futures.emplace_back(pool(policy).submit([&f](InputIt chunk_first, InputIt chunk_last) {
+                for (; chunk_first != chunk_last; ++chunk_first) {
+                    f(*chunk_first);
                 }
             }, first, loop_end));
 

include/poolstl/execution

@@ -41,26 +41,24 @@ namespace poolstl {
         }
 
         template <typename ExecutionPolicy>
-        ttp::task_thread_pool& pool(ExecutionPolicy& policy) {
-            if constexpr (ExecutionPolicy::use_default_pool()) {
-                return *get_default_pool();
-            } else {
-                return policy.pool;
-            }
+        ttp::task_thread_pool& pool(ExecutionPolicy&) {
+            return *get_default_pool();
+        }
+
+        template <>
+        inline ttp::task_thread_pool& pool<par_pool>(par_pool& policy) {
+            return policy.pool;
         }
 
         template <typename T> struct is_poolstl_execution_policy : std::false_type {};
         template <> struct is_poolstl_execution_policy<::poolstl::execution::parallel_policy> : std::true_type {};
         template <> struct is_poolstl_execution_policy<::poolstl::execution::par_pool> : std::true_type {};
 
-        template <class T>
-        inline constexpr bool is_poolstl_execution_policy_v = is_poolstl_execution_policy<T>::value;
-
         template <class ExecutionPolicy, class Tp>
         using enable_if_poolstl_execution_policy =
             typename std::enable_if<
-                is_poolstl_execution_policy_v<
-                    typename std::remove_cv<typename std::remove_reference<ExecutionPolicy>::type>::type>,
+                is_poolstl_execution_policy<
+                    typename std::remove_cv<typename std::remove_reference<ExecutionPolicy>::type>::type>::value,
                 Tp>::type;
     }
 }

include/poolstl/internal/utils.hpp

@@ -12,16 +12,24 @@
 #define POOLSTL_VERSION_PATCH 0
 
 #include <cstddef>
+#include <iterator>
 
-namespace poolstl::internal {
+namespace poolstl {
+    namespace internal {
 
-    template<typename Iterator>
-    std::size_t get_chunk_size(Iterator first, Iterator last, unsigned int num_threads) {
-        std::size_t num_steps = (last - first);
-        auto remainder = num_steps % num_threads;
-        return (num_steps / num_threads) + (remainder > 0 ? 1 : 0);
-    }
+        inline std::size_t get_chunk_size(std::size_t num_steps, unsigned int num_threads) {
+            auto remainder = num_steps % num_threads;
+            return (num_steps / num_threads) + (remainder > 0 ? 1 : 0);
+        }
 
+        template<typename Iterator>
+        typename std::iterator_traits<Iterator>::difference_type
+        get_chunk_size(Iterator first, Iterator last, unsigned int num_threads) {
+            using diff_t = typename std::iterator_traits<Iterator>::difference_type;
+            std::size_t num_steps = std::distance(first, last);
+            return static_cast<diff_t>(get_chunk_size(num_steps, num_threads));
+        }
+    }
 }
 
 #endif

tests/CMakeLists.txt

@@ -22,7 +22,7 @@ Include(FetchContent)
 FetchContent_Declare(
         Catch2
         GIT_REPOSITORY https://github.com/catchorg/Catch2.git
-        GIT_TAG        v3.3.2
+        GIT_TAG        v3.4.0
         GIT_SHALLOW TRUE
         EXCLUDE_FROM_ALL
 )
@@ -35,4 +35,13 @@ include(Catch)
 
 add_executable(poolstl_test poolstl_test.cpp)
 target_link_libraries(poolstl_test PRIVATE Catch2::Catch2WithMain poolSTL::poolSTL)
+target_compile_definitions(poolstl_test PRIVATE CATCH_CONFIG_FAST_COMPILE)
+# Catch2 requires C++14
+target_compile_features(poolstl_test PUBLIC cxx_std_14)
+
 catch_discover_tests(poolstl_test)
+
+# Dedicated target to ensure C++11 builds work
+add_executable(cpp11_test cpp11_test.cpp)
+target_link_libraries(cpp11_test PUBLIC poolSTL::poolSTL)
+target_compile_features(cpp11_test PUBLIC cxx_std_11)

tests/cpp11_test.cpp

@@ -0,0 +1,17 @@
+// Copyright (C) 2023 Adam Lugowski. All rights reserved.
+// Use of this source code is governed by:
+// the BSD 2-clause license, the MIT license, or at your choosing the BSL-1.0 license found in the LICENSE.*.txt files.
+// SPDX-License-Identifier: BSD-2-Clause OR MIT OR BSL-1.0
+
+#include <iostream>
+
+#include <poolstl/poolstl.hpp>
+
+int main() {
+    std::vector<int> v = {0, 1, 2, 3, 4, 5};
+    std::for_each(poolstl::par, v.cbegin(), v.cend(), [](int x) {
+        std::cout << x << " ";
+    });
+    std::cout << std::endl;
+    return 0;
+}

tests/poolstl_test.cpp

@@ -13,7 +13,6 @@
 #include "utils.hpp"
 
 namespace ttp = task_thread_pool;
-using it = i_iter<int>;
 
 TEST_CASE("for_each", "[alg]") {
     std::atomic<int> sum{0};
@@ -40,8 +39,9 @@ TEST_CASE("for_each", "[alg]") {
 TEST_CASE("default_pool", "[execution]") {
     std::atomic<int> sum{0};
     for (auto num_iters : test_arr_sizes) {
+        auto v = iota_vector(num_iters);
         sum = 0;
-        std::for_each(poolstl::par, it(0), it(num_iters), [&](auto) { ++sum; });
+        std::for_each(poolstl::par, v.cbegin(), v.cend(), [&](auto) { ++sum; });
         REQUIRE(sum == num_iters);
     }
 }

tests/utils.hpp

@@ -10,26 +10,6 @@
 constexpr std::array<int, 6> test_thread_counts = {0, 1, 2, 3, 5, 10};
 constexpr std::array<int, 14> test_arr_sizes = {0, 1, 2, 3, 4, 8, 9, 10, 11, 20, 23, 77, 101};
 
-template<typename T>
-struct i_iter {
-    T value;
-
-    explicit i_iter(T v) : value(v) {}
-
-    T operator*() const { return value; }
-
-    T operator++() { return ++value; }
-
-    template<typename U>
-    i_iter operator+(const U &other) { return i_iter<T>(value + other); }
-
-    T operator-(const i_iter<T> &other) const { return value - other.value; }
-
-    bool operator<(const i_iter<T> &other) const { return value < other.value; }
-    bool operator==(const i_iter<T> &other) const { return value == other.value; }
-    bool operator!=(const i_iter<T> &other) const { return value != other.value; }
-};
-
 template <typename T=int>
 std::vector<T> iota_vector(size_t size, T init=0) {
     std::vector<T> ret(size);