Merge pull request #427 from eseiler/fix/codechecker

[FIX] static analysis

.github/workflows/ci_linux.yml

@@ -33,7 +33,7 @@ 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"]
         include:
           - cxx_flags: "-Wno-interference-size"
           - compiler: "clang-18"

.github/workflows/ci_macos.yml

@@ -33,7 +33,7 @@ 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"]
         include:
           - cxx_flags: "-Wno-interference-size"
           - compiler: "clang-18"

.github/workflows/ci_misc.yml

@@ -33,7 +33,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        compiler: ["clang-18", "gcc-14", "gcc-11", "intel"]
+        compiler: ["clang-18", "gcc-14", "gcc-12", "intel"]
         build: ["snippet", "performance", "header"]
         include:
           - cxx_flags: "-Wno-interference-size"

.github/workflows/ci_utility.yml

@@ -33,7 +33,7 @@ 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"]
         include:
           - cxx_flags: "-Wno-interference-size"
           - compiler: "clang-18"

doc/setup/index.md

@@ -20,7 +20,8 @@ conda install -c bioconda -c conda-forge raptor
 
 ### Prerequisites
 * CMake >= 3.21
-* GCC 11, 12 or 13 (most recent minor version)
+* GCC 12, 13 or 14 (most recent minor version)
+* Clang 17 or 18 (most recent minor version)
 * git
 
 Refer to the [Seqan3 Setup Tutorial](https://docs.seqan.de/seqan/3-master-user/setup.html) for more in depth

include/raptor/build/emplace_iterator.hpp

@@ -32,13 +32,13 @@ class emplace_iterator
 
     explicit constexpr emplace_iterator(seqan::hibf::interleaved_bloom_filter & ibf, seqan::hibf::bin_index const idx) :
         ibf{std::addressof(ibf)},
-        index{std::move(idx)}
+        index{idx}
     {}
 
     /* constexpr */ emplace_iterator & operator=(uint64_t const value) noexcept
     {
         assert(ibf != nullptr);
-        ibf->emplace(std::move(value), index);
+        ibf->emplace(value, index);
         return *this;
     }
 
@@ -65,7 +65,7 @@ class emplace_iterator
 [[nodiscard]] inline constexpr emplace_iterator emplacer(seqan::hibf::interleaved_bloom_filter & ibf,
                                                          seqan::hibf::bin_index const idx)
 {
-    return emplace_iterator{ibf, std::move(idx)};
+    return emplace_iterator{ibf, idx};
 }
 
 } // namespace raptor

include/raptor/threshold/forward_strand_minimiser.hpp

@@ -94,38 +94,45 @@ struct forward_strand_minimiser
         auto kmer_view = text | seqan3::views::kmer_hash(shape) | std::views::transform(apply_xor);
         forward_hashes.assign(kmer_view.begin(), kmer_view.end());
 
+        struct kmer
+        {
+            uint64_t hash{};
+            uint64_t position{};
+
+            constexpr auto operator<=>(kmer const & other) const = default;
+        };
+
         // Stores hash and begin for all k-mers in the window
-        std::deque<std::pair<uint64_t, uint64_t>> window_hashes;
+        std::deque<kmer> window_hashes;
 
         // Initialisation. We need to compute all hashes for the first window.
         for (uint64_t i = 0; i < kmers_per_window; ++i)
-            window_hashes.emplace_back(forward_hashes[i], i);
+            window_hashes.emplace_back(kmer{.hash = forward_hashes[i], .position = i});
 
         // The minimum hash is the minimiser. Store the begin position.
-        auto min = std::min_element(std::begin(window_hashes), std::end(window_hashes));
-        minimiser_begin.push_back(min->second);
+        auto min = std::ranges::min(window_hashes);
+        minimiser_begin.push_back(min.position);
 
         // For the following windows, we remove the first window k-mer (is now not in window) and add the new k-mer
         // that results from the window shifting.
         for (uint64_t i = kmers_per_window; i < max_number_of_minimiser; ++i)
         {
-            // Already store the new hash without removing the first one.
             uint64_t const new_hash{forward_hashes[i + kmers_per_window - 1]}; // Already did kmers_per_window - 1 many
-            window_hashes.emplace_back(new_hash, i);
 
-            if (new_hash < min->second) // New hash is the minimum.
-            {
-                min = std::prev(std::end(window_hashes));
-                minimiser_begin.push_back(min->second);
-            }
-            else if (min == std::begin(window_hashes)) // Minimum is the yet-to-be-removed begin of the window.
+            // There are two conditions when we need to recompute the minimum:
+            bool const minimiser_leaves_window = window_hashes.front() == min;
+            bool const new_hash_is_min = new_hash < min.hash;
+
+            // Rolling hash / sliding window
+            window_hashes.pop_front();
+            window_hashes.emplace_back(kmer{.hash = new_hash, .position = i});
+
+            // Update the minimum
+            if (new_hash_is_min || minimiser_leaves_window)
             {
-                // The first hash will be removed, the last one is caught by the previous if.
-                min = std::min_element(++std::begin(window_hashes), std::prev(std::end(window_hashes)));
-                minimiser_begin.push_back(min->second);
+                min = new_hash_is_min ? window_hashes.back() : std::ranges::min(window_hashes);
+                minimiser_begin.push_back(min.position);
             }
-
-            window_hashes.pop_front(); // Remove the first k-mer.
         }
         return;
     }

test/include/raptor/test/cli_test.hpp

@@ -384,18 +384,7 @@ struct raptor_base : public cli_test
 
             for (auto && hit : std::views::split(line_view.substr(query_prefix.size() + 2u), ','))
             {
-// GCC 11 does not implement P2210R2, hence GCC 11's split_view is actually a lazy_split_view.
-#if defined(__GNUC__) && !defined(__llvm__) && !defined(__INTEL_COMPILER) && (__GNUC__ > 10) && (__GNUC__ < 12)
-                std::stringstream buf;
-                for (auto const & inner_view : hit)
-                {
-                    buf << inner_view;
-                }
-                auto view = buf.view();
-                std::from_chars(view.data(), view.data() + view.size(), tmp);
-#else
                 std::from_chars(hit.data(), hit.data() + hit.size(), tmp);
-#endif
                 actual_hits.push_back(tmp);
             }
             std::ranges::sort(actual_hits);

test/unit/api/threshold.cpp

@@ -116,11 +116,11 @@ TEST(minimiser, logspace_substract)
     raptor::threshold::threshold const threshold{parameters};
     // results for pseudorandom generators are implementation-defined
 #ifdef _LIBCPP_VERSION
-    std::vector<size_t> const expected{1u,  1u,  1u,  1u,  2u,  3u,  4u,  5u,  6u,  7u,  8u,  8u, 9u,
-                                       10u, 11u, 12u, 13u, 14u, 15u, 16u, 17u, 18u, 19u, 21u, 22u};
+    std::vector<size_t> const expected{1u,  1u,  1u,  1u,  2u,  4u,  5u,  6u,  7u,  8u,  9u,  9u, 10u,
+                                       11u, 12u, 13u, 14u, 15u, 16u, 17u, 18u, 19u, 20u, 22u, 23u};
 #else
-    std::vector<size_t> const expected{1u,  1u,  1u,  1u,  2u,  3u,  4u,  5u,  6u,  6u,  7u,  8u, 9u,
-                                       10u, 11u, 12u, 13u, 13u, 14u, 15u, 16u, 17u, 18u, 20u, 21u};
+    std::vector<size_t> const expected{1u,  1u,  1u,  1u,  2u,  4u,  5u,  6u,  7u,  7u,  8u,  9u, 10u,
+                                       11u, 12u, 13u, 14u, 15u, 16u, 17u, 18u, 19u, 20u, 22u, 23u};
 #endif
     ASSERT_EQ(expected.size(), 37u - 12u);
     for (size_t i = 12u; i < 37u; ++i)