[MISC] CPP23 md subscript

Author: eseiler

.github/workflows/ci_linux.yml

@@ -32,7 +32,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        compiler: ["clang-18", "clang-17", "gcc-14", "gcc-13", "gcc-12", "gcc-11", "intel"]
+        compiler: ["clang-18", "clang-17", "gcc-14", "gcc-13", "gcc-12", "intel"]
     container:
       image: ghcr.io/seqan/${{ matrix.compiler }}
       volumes:

.github/workflows/ci_macos.yml

@@ -32,7 +32,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        compiler: ["clang-18", "clang-17", "gcc-14", "gcc-13", "gcc-12", "gcc-11"]
+        compiler: ["clang-18", "clang-17", "gcc-14", "gcc-13", "gcc-12"]
     steps:
       - name: Checkout
         uses: actions/checkout@v4

.github/workflows/ci_misc.yml

@@ -40,8 +40,8 @@ jobs:
             use_include_dependencies: "OFF"
             cxx_flags: "-stdlib=libc++"
 
-          - name: "Snippet gcc11"
-            compiler: "gcc-11"
+          - name: "Snippet gcc12"
+            compiler: "gcc-12"
             build: snippet
             build_type: Release
             use_include_dependencies: "OFF"
@@ -53,8 +53,8 @@ jobs:
             use_include_dependencies: "OFF"
             cxx_flags: "-stdlib=libc++"
 
-          - name: "Performance gcc11"
-            compiler: "gcc-11"
+          - name: "Performance gcc12"
+            compiler: "gcc-12"
             build: performance
             build_type: Release
             use_include_dependencies: "OFF"
@@ -72,8 +72,8 @@ jobs:
             build_type: Release
             use_include_dependencies: "OFF"
 
-          - name: "Header gcc11"
-            compiler: "gcc-11"
+          - name: "Header gcc12"
+            compiler: "gcc-12"
             build: header
             build_type: Release
             use_include_dependencies: "OFF"
@@ -85,7 +85,7 @@ jobs:
             use_include_dependencies: "ON"
 
           - name: "Tutorial"
-            compiler: "gcc-11"
+            compiler: "gcc-12"
             build: tutorial
             build_type: Debug
             use_include_dependencies: "OFF"

cmake/configuration.cmake

@@ -79,43 +79,23 @@ include ("${HIBF_SOURCE_DIR}/test/cmake/hibf_require_ccache.cmake")
 hibf_require_ccache ()
 
 # ----------------------------------------------------------------------------
-# Require C++20
+# C++ standard
 # ----------------------------------------------------------------------------
 
-set (CMAKE_REQUIRED_FLAGS_SAVE ${CMAKE_REQUIRED_FLAGS})
-
-set (CXXSTD_TEST_SOURCE
-     "#if !defined (__cplusplus) || (__cplusplus < 202002)
-      #error NOCXX20
-      #endif
-      int main() {}")
-
-set (HIBF_FEATURE_CPP20_FLAG_BUILTIN "")
-set (HIBF_FEATURE_CPP20_FLAG_STD20 "-std=c++20")
-set (HIBF_FEATURE_CPP20_FLAG_STD2a "-std=c++2a")
-
-set (HIBF_CPP20_FLAG "")
-
-foreach (_FLAG BUILTIN STD20 STD2a)
-    set (CMAKE_REQUIRED_FLAGS "${CMAKE_REQUIRED_FLAGS_SAVE} ${HIBF_FEATURE_CPP20_FLAG_${_FLAG}}")
-
-    check_cxx_source_compiles ("${CXXSTD_TEST_SOURCE}" CPP20_FLAG_${_FLAG})
+if (NOT DEFINED CMAKE_CXX_STANDARD_REQUIRED)
+    set (CMAKE_CXX_STANDARD_REQUIRED OFF)
+endif ()
 
-    if (CPP20_FLAG_${_FLAG})
-        set (HIBF_CPP20_FLAG ${_FLAG})
-        break ()
-    endif ()
-endforeach ()
+if (NOT DEFINED CMAKE_CXX_EXTENSIONS)
+    set (CMAKE_CXX_EXTENSIONS OFF)
+endif ()
 
-set (CMAKE_REQUIRED_FLAGS ${CMAKE_REQUIRED_FLAGS_SAVE})
+if (NOT DEFINED CMAKE_CXX_STANDARD)
+    set (CMAKE_CXX_STANDARD 23)
+endif ()
 
-if (HIBF_CPP20_FLAG STREQUAL "BUILTIN")
-    hibf_config_print ("C++ Standard-20 support:    builtin")
-elseif (HIBF_CPP20_FLAG)
-    set (HIBF_CXX_FLAGS "${HIBF_CXX_FLAGS} ${HIBF_FEATURE_CPP20_FLAG_${HIBF_CPP20_FLAG}}")
-    hibf_config_print ("C++ Standard-20 support:    via ${HIBF_FEATURE_CPP20_FLAG_${HIBF_CPP20_FLAG}}")
-else ()
-    hibf_config_error ("HIBF requires C++20, but your compiler does not support it.")
+if (DEFINED CMAKE_CXX_STANDARD AND CMAKE_CXX_STANDARD LESS 23)
+    message (FATAL_ERROR "Requires C++23 or higher.")
 endif ()
 
 # ----------------------------------------------------------------------------

include/hibf/hierarchical_interleaved_bloom_filter.hpp

@@ -21,8 +21,9 @@
 #include <hibf/interleaved_bloom_filter.hpp> // for interleaved_bloom_filter
 #include <hibf/layout/layout.hpp>            // for layout
 #include <hibf/misc/counting_vector.hpp>     // for counting_vector
-#include <hibf/misc/timer.hpp>               // for concurrent_timer
-#include <hibf/platform.hpp>                 // for HIBF_CONSTEXPR_VECTOR
+#include <hibf/misc/md_vector.hpp>
+#include <hibf/misc/timer.hpp> // for concurrent_timer
+#include <hibf/platform.hpp>   // for HIBF_CONSTEXPR_VECTOR
 
 namespace seqan::hibf
 {
@@ -202,7 +203,7 @@ class hierarchical_interleaved_bloom_filter
      * If `j != i` is returned, there is a lower level IBF, bin `b` is a merged bin, and `j` is the ID of the lower
      * level IBF in ibf_vector.
      */
-    std::vector<std::vector<int64_t>> next_ibf_id;
+    md_vector<int64_t> next_ibf_id;
 
     /*!\brief Stores for each bin in each IBF of the HIBF the user bin ID.
     * \details
@@ -211,7 +212,7 @@ class hierarchical_interleaved_bloom_filter
     * lower level IBF.
     * Otherwise, the returned value `j` is the corresponding user bin ID.
     */
-    std::vector<std::vector<int64_t>> ibf_bin_to_user_bin_id{};
+    md_vector<int64_t> ibf_bin_to_user_bin_id{};
 
     //!\brief Returns a membership_agent to be used for counting.
     membership_agent_type membership_agent() const;
@@ -280,16 +281,16 @@ class hierarchical_interleaved_bloom_filter::membership_agent_type
         {
             sum += result[bin];
 
-            auto const current_filename_index = hibf_ptr->ibf_bin_to_user_bin_id[ibf_idx][bin];
+            auto const current_filename_index = hibf_ptr->ibf_bin_to_user_bin_id[ibf_idx, bin];
 
             if (current_filename_index < 0) // merged bin
             {
                 if (sum >= threshold)
-                    membership_for_impl(values, hibf_ptr->next_ibf_id[ibf_idx][bin], threshold);
+                    membership_for_impl(values, hibf_ptr->next_ibf_id[ibf_idx, bin], threshold);
                 sum = 0u;
             }
             else if (bin + 1u == result.size() ||                                                  // last bin
-                     current_filename_index != hibf_ptr->ibf_bin_to_user_bin_id[ibf_idx][bin + 1]) // end of split bin
+                     current_filename_index != hibf_ptr->ibf_bin_to_user_bin_id[ibf_idx, bin + 1]) // end of split bin
             {
                 if (sum >= threshold)
                     result_buffer.emplace_back(current_filename_index);
@@ -415,16 +416,16 @@ class hierarchical_interleaved_bloom_filter::counting_agent_type
         for (size_t bin{}; bin < result.size(); ++bin)
         {
             sum += result[bin];
-            auto const current_filename_index = hibf_ptr->ibf_bin_to_user_bin_id[ibf_idx][bin];
+            auto const current_filename_index = hibf_ptr->ibf_bin_to_user_bin_id[ibf_idx, bin];
 
             if (current_filename_index < 0) // merged bin
             {
                 if (sum >= threshold)
-                    bulk_count_impl(values, hibf_ptr->next_ibf_id[ibf_idx][bin], threshold);
+                    bulk_count_impl(values, hibf_ptr->next_ibf_id[ibf_idx, bin], threshold);
                 sum = 0u;
             }
             else if (bin + 1u == result.size() ||                                                  // last bin
-                     current_filename_index != hibf_ptr->ibf_bin_to_user_bin_id[ibf_idx][bin + 1]) // end of split bin
+                     current_filename_index != hibf_ptr->ibf_bin_to_user_bin_id[ibf_idx, bin + 1]) // end of split bin
             {
                 if (sum >= threshold)
                     result_buffer[current_filename_index] = sum;

include/hibf/layout/hierarchical_binning.hpp

@@ -13,7 +13,8 @@
 #include <hibf/build/bin_size_in_bits.hpp> // for bin_size_in_bits
 #include <hibf/config.hpp>                 // for config
 #include <hibf/layout/data_store.hpp>      // for data_store
-#include <hibf/platform.hpp>               // for HIBF_WORKAROUND_GCC_BOGUS_MEMCPY
+#include <hibf/misc/md_vector.hpp>
+#include <hibf/platform.hpp> // for HIBF_WORKAROUND_GCC_BOGUS_MEMCPY
 
 namespace seqan::hibf::layout
 {
@@ -140,9 +141,9 @@ class hierarchical_binning
      *
      * \image html hierarchical_dp_init.png
      */
-    void initialization(std::vector<std::vector<size_t>> & matrix,
-                        std::vector<std::vector<size_t>> & ll_matrix,
-                        std::vector<std::vector<std::pair<size_t, size_t>>> & trace);
+    void initialization(md_vector<size_t> & matrix,
+                        md_vector<size_t> & ll_matrix,
+                        md_vector<std::pair<size_t, size_t>> & trace);
 
     /*!\brief Performs the recursion.
      *
@@ -182,9 +183,8 @@ class hierarchical_binning
      * this algorithm. It would be too computational intensive to compute the splitting for every possibility.
      *
      */
-    void recursion(std::vector<std::vector<size_t>> & matrix,
-                   std::vector<std::vector<size_t>> & ll_matrix,
-                   std::vector<std::vector<std::pair<size_t, size_t>>> & trace);
+    void
+    recursion(md_vector<size_t> & matrix, md_vector<size_t> & ll_matrix, md_vector<std::pair<size_t, size_t>> & trace);
 
     void backtrack_merged_bin(size_t trace_j,
                               size_t const next_j,
@@ -198,7 +198,7 @@ class hierarchical_binning
                              maximum_bin_tracker & max_tracker);
 
     //!\brief Backtracks the trace matrix and writes the resulting binning into the output file.
-    size_t backtracking(std::vector<std::vector<std::pair<size_t, size_t>>> const & trace);
+    size_t backtracking(md_vector<std::pair<size_t, size_t>> const & trace);
 
     data_store initialise_libf_data(size_t const trace_j) const;
 

include/hibf/misc/md_vector.hpp

@@ -0,0 +1,42 @@
+// SPDX-FileCopyrightText: 2006-2024, Knut Reinert & Freie Universität Berlin
+// SPDX-FileCopyrightText: 2016-2024, Knut Reinert & MPI für molekulare Genetik
+// SPDX-License-Identifier: BSD-3-Clause
+
+/*!\file
+ * \brief Provides seqan::hibf::md_vector.
+ * \author Enrico Seiler <enrico.seiler AT fu-berlin.de>
+ */
+
+#pragma once
+
+#include <vector> // for vector
+
+#include <hibf/platform.hpp>
+
+namespace seqan::hibf
+{
+
+template <typename value_t>
+struct md_vector : public std::vector<std::vector<value_t>>
+{
+    using base_t = std::vector<std::vector<value_t>>;
+    using base_t::base_t;
+    using base_t::operator[];
+#if defined(__cpp_explicit_this_parameter) && __cpp_explicit_this_parameter >= 202110L
+    decltype(auto) operator[](this auto & self, size_t const x, size_t const y)
+    {
+        return self[x][y];
+    }
+#else
+    value_t & operator[](size_t const x, size_t const y)
+    {
+        return (*this)[x][y];
+    }
+    value_t const & operator[](size_t const x, size_t const y) const
+    {
+        return (*this)[x][y];
+    }
+#endif
+};
+
+} // namespace seqan::hibf

include/hibf/sketch/minhashes.hpp

@@ -17,6 +17,7 @@
 #include <cereal/access.hpp> // for access
 #include <cereal/cereal.hpp> // for make_nvp, CEREAL_NVP
 
+#include <hibf/misc/md_vector.hpp>
 #include <hibf/platform.hpp>
 
 namespace seqan::hibf::sketch
@@ -39,7 +40,7 @@ struct minhashes
     static constexpr size_t sketch_size{40};
 
     //!\brief A table of sketches. For LSH we need multiple sketches, stored in a table.
-    std::vector<std::vector<uint64_t>> table{}; // Each element (vector<uint64_t>) is a minhash.
+    md_vector<uint64_t> table{}; // Each element (vector<uint64_t>) is a minhash.
 
     /*!\name Constructors, destructor and assignment
      * \{