improvement

Add missing `find_dependency` calls.

.github/workflows/cmake-multi-platform.yml

@@ -245,72 +245,37 @@ jobs:
         run: |
           ./out/Debug/unit_tests.exe
 
-      - name: Process Coverage (llvm-profdata & llvm-cov)
+      - name: Process and render coverage
         if: ${{ matrix.triplet == 'x64-windows' }}
         shell: pwsh
         run: |
-          $BUILD_DIR = "cmake-build/build/${{ matrix.preset }}"
-          $EXE_PATH = "./out/Debug/unit_tests.exe"
-          
-          # 1. Merge raw profile data (essential step)
+          $buildDir = "cmake-build/build/${{ matrix.preset }}"
+          $exePath = "./out/Debug/unit_tests.exe"
+          $profile = "$buildDir/unit_tests.profdata"
+          $htmlDir = "$buildDir/coverage-html"
+
           & "C:/Program Files/LLVM/bin/llvm-profdata.exe" merge `
-            -sparse "$BUILD_DIR/unit_tests.profraw" `
-            -o "$BUILD_DIR/unit_tests.profdata"
+            -sparse "$buildDir/unit_tests.profraw" `
+            -o $profile
 
-          # 2. Export to LCOV format
-          #    NOTE: We explicitly ignore vcpkg_installed and system headers
-          & "C:/Program Files/LLVM/bin/llvm-cov.exe" export "$EXE_PATH" `
-            -instr-profile="$BUILD_DIR/unit_tests.profdata" `
-            -format=lcov `
-            -ignore-filename-regex="vcpkg_installed|external|tests" `
-            > "$BUILD_DIR/lcov.info"
-
-          if (Test-Path "$BUILD_DIR/lcov.info") {
-            Write-Host "✅ LCOV info created at $BUILD_DIR/lcov.info"
-          } else {
-            Write-Error "Failed to create LCOV info"
-            exit 1
+          if ($LASTEXITCODE -ne 0) {
+            exit $LASTEXITCODE
           }
 
-      - name: Install LCOV (for genhtml)
-        if: ${{ matrix.triplet == 'x64-windows' }}
-        run: choco install lcov -y
+          & "C:/Program Files/LLVM/bin/llvm-cov.exe" show $exePath `
+            "-instr-profile=$profile" `
+            "-format=html" `
+            "-output-dir=$htmlDir" `
+            "-ignore-filename-regex=vcpkg_installed|external|tests" `
+            "-show-branches=count"
 
-      - name: Generate HTML Report
-        if: ${{ matrix.triplet == 'x64-windows' }}
-        shell: bash
-        run: |
-          BUILD_DIR="cmake-build/build/${{ matrix.preset }}"
-          LCOV_INFO="${BUILD_DIR}/lcov.info"
-          HTML_DIR="${BUILD_DIR}/coverage-html"
+          if ($LASTEXITCODE -ne 0) {
+            exit $LASTEXITCODE
+          }
 
-          # Fix paths for genhtml (Perl hates backslashes)
-          sed -i 's|\\|/|g' "${LCOV_INFO}"
-          
-          # Locate genhtml provided by 'choco install lcov'
-          # It is typically in ProgramData/chocolatey/lib/lcov/tools/bin
-          GENHTML=$(find /c/ProgramData/chocolatey -name genhtml -print -quit)
-          
-          if [ -z "$GENHTML" ]; then
-            echo "Error: genhtml executable not found"
-            exit 1
-          fi
-          
-          echo "Using genhtml: $GENHTML"
-          mkdir -p "$HTML_DIR"
-          
-          # Run genhtml
-          # Added --demangle-cpp if your version supports it, otherwise remove it
-          perl "$GENHTML" \
-            "${LCOV_INFO}" \
-            --output-directory "$HTML_DIR" \
-            --title "OMath Coverage Report" \
-            --legend \
-            --show-details \
-            --branch-coverage \
-            --ignore-errors source
-            
-          echo "✅ LCOV HTML report generated at $HTML_DIR"
+          if (-not (Test-Path "$htmlDir/index.html")) {
+            throw "LLVM coverage report was not generated"
+          }
 
       - name: Upload Coverage (HTML Report)
         if: ${{ matrix.triplet == 'x64-windows' }}
@@ -742,7 +707,11 @@ jobs:
 
       - name: Build
         run: |
-          cmake --build cmake-build/build/${{ matrix.preset }} --target unit_tests omath
+          if [[ "${{ matrix.msystem }}" == "MINGW32" ]]; then
+            cmake --build cmake-build/build/${{ matrix.preset }} --target unit_tests omath --parallel 1
+          else
+            cmake --build cmake-build/build/${{ matrix.preset }} --target unit_tests omath
+          fi
 
       - name: Run unit_tests.exe
         run: |

VERSION

@@ -1 +1 @@
-5.4.0
+5.5.1

benchmark/benchmark_mat.cpp

@@ -2,11 +2,9 @@
 // Created by Vlad on 9/17/2025.
 //
 #include <benchmark/benchmark.h>
-
 #include <omath/omath.hpp>
 using namespace omath;
 
-
 void mat_float_multiplication_col_major(benchmark::State& state)
 {
     using MatType = Mat<128, 128, float, MatStoreType::COLUMN_MAJOR>;
@@ -15,9 +13,12 @@ void mat_float_multiplication_col_major(benchmark::State& state)
     a.set(3.f);
     b.set(7.f);
 
-
     for ([[maybe_unused]] const auto _ : state)
-        std::ignore = a * b;
+    {
+        benchmark::DoNotOptimize(a);
+        benchmark::DoNotOptimize(b);
+        benchmark::DoNotOptimize(a * b);
+    }
 }
 void mat_float_multiplication_row_major(benchmark::State& state)
 {
@@ -27,9 +28,12 @@ void mat_float_multiplication_row_major(benchmark::State& state)
     a.set(3.f);
     b.set(7.f);
 
-
     for ([[maybe_unused]] const auto _ : state)
-        std::ignore = a * b;
+    {
+        benchmark::DoNotOptimize(a);
+        benchmark::DoNotOptimize(b);
+        benchmark::DoNotOptimize(a * b);
+    }
 }
 
 void mat_double_multiplication_row_major(benchmark::State& state)
@@ -40,9 +44,12 @@ void mat_double_multiplication_row_major(benchmark::State& state)
     a.set(3.f);
     b.set(7.f);
 
-
     for ([[maybe_unused]] const auto _ : state)
-        std::ignore = a * b;
+    {
+        benchmark::DoNotOptimize(a);
+        benchmark::DoNotOptimize(b);
+        benchmark::DoNotOptimize(a * b);
+    }
 }
 
 void mat_double_multiplication_col_major(benchmark::State& state)
@@ -53,13 +60,16 @@ void mat_double_multiplication_col_major(benchmark::State& state)
     a.set(3.f);
     b.set(7.f);
 
-
     for ([[maybe_unused]] const auto _ : state)
-        std::ignore = a * b;
+    {
+        benchmark::DoNotOptimize(a);
+        benchmark::DoNotOptimize(b);
+        benchmark::DoNotOptimize(a * b);
+    }
 }
 
 BENCHMARK(mat_float_multiplication_col_major)->Iterations(5000);
 BENCHMARK(mat_float_multiplication_row_major)->Iterations(5000);
 
 BENCHMARK(mat_double_multiplication_col_major)->Iterations(5000);
-BENCHMARK(mat_double_multiplication_row_major)->Iterations(5000);
+BENCHMARK(mat_double_multiplication_row_major)->Iterations(5000);

cmake/omathConfig.cmake.in

@@ -6,6 +6,17 @@ if (@OMATH_IMGUI_INTEGRATION@)
     find_dependency(imgui CONFIG)
 endif()
 
+if (@OMATH_ENABLE_LUA@)
+    find_dependency(Lua)
+endif()
+
+if (@OMATH_ENABLE_HOOKING@)
+    find_dependency(safetyhook CONFIG)
+    if (NOT WIN32 AND (UNIX AND NOT APPLE))
+        find_dependency(OpenGL)
+    endif()
+endif()
+
 # Load the targets for the omath library
 include("${CMAKE_CURRENT_LIST_DIR}/omathTargets.cmake")
 check_required_components(omath)

docs/trigonometry/angle.md

@@ -3,8 +3,8 @@
 > Header: `omath/trigonometry/angle.hpp`
 > Namespace: `omath`
 > Template: `Angle<Type = float, min = 0, max = 360, flags = AngleFlags::Normalized>`
-> Requires: `std::is_arithmetic_v<Type>`
-> Formatters: `std::formatter` for `char`, `wchar_t`, `char8_t` → `"{}deg"`
+> Requires: `std::is_floating_point_v<Type>`
+> Formatters: `std::formatter` for `char` and `wchar_t` → `"{}deg"`
 
 ---
 
@@ -14,7 +14,7 @@
 
 Two behaviors via `AngleFlags`:
 
-* `AngleFlags::Normalized` (default): values are wrapped into `[min, max]` using `angles::wrap_angle`.
+* `AngleFlags::Normalized` (default): values are wrapped into `[min, max)` using `angles::wrap_angle`.
 * `AngleFlags::Clamped`: values are clamped to `[min, max]` using `std::clamp`.
 
 ---
@@ -28,12 +28,16 @@ enum class AngleFlags { Normalized = 0, Clamped = 1 };
 
 template<class Type = float, Type min = Type(0), Type max = Type(360),
          AngleFlags flags = AngleFlags::Normalized>
-requires std::is_arithmetic_v<Type>
+requires std::is_floating_point_v<Type>
 class Angle {
 public:
   // Construction
   static constexpr Angle from_degrees(const Type& deg) noexcept;
   static constexpr Angle from_radians(const Type& rad) noexcept;
+  static constexpr Angle from_asin(const Type& value) noexcept;
+  static constexpr Angle from_acos(const Type& value) noexcept;
+  static constexpr Angle from_atan(const Type& value) noexcept;
+  static constexpr Angle from_atan2(const Type& y, const Type& x) noexcept;
   constexpr Angle() noexcept;                  // 0 deg, adjusted by flags/range
 
   // Accessors / conversions (degrees stored internally)
@@ -45,10 +49,9 @@ public:
   Type sin() const noexcept;
   Type cos() const noexcept;
   Type tan() const noexcept;
-  Type atan() const noexcept;  // atan(as_radians())  (rarely used)
   Type cot() const noexcept;   // cos()/sin()  (watch sin≈0)
 
-  // Arithmetic (wraps or clamps per flags and [min,max])
+  // Arithmetic (wraps or clamps per flags and configured range)
   constexpr Angle& operator+=(const Angle&) noexcept;
   constexpr Angle& operator-=(const Angle&) noexcept;
   constexpr Angle  operator+(const Angle&) noexcept;
@@ -68,7 +71,7 @@ public:
 std::format("{}", Angle<float>::from_degrees(45)); // "45deg"
 ```
 
-Formatters exist for `char`, `wchar_t`, and `char8_t`.
+Formatters exist for `char` and `wchar_t`.
 
 ---
 
@@ -116,10 +119,9 @@ float deg = *yaw;                                // same as yaw.as_degrees()
 ## Semantics & notes
 
 * **Storage & units:** Internally stores **degrees** (`Type m_angle`). `as_radians()`/`from_radians()` use the project helpers in `omath::angles`.
-* **Arithmetic honors policy:** `operator+=`/`-=` and the binary `+`/`-` apply **wrap** or **clamp** in `[min,max]`, mirroring construction behavior.
-* **`atan()`**: returns `std::atan(as_radians())` (the arctangent of the *radian value*). This is mathematically unusual for an angle type and is rarely useful; prefer `tan()`/`atan2` in client code when solving geometry problems.
+* **Arithmetic honors policy:** `operator+=`/`-=` and the binary `+`/`-` apply **wrap** or **clamp**, mirroring construction behavior.
 * **`cot()` / `tan()` singularities:** Near multiples where `sin() ≈ 0` or `cos() ≈ 0`, results blow up. Guard in your usage if inputs can approach these points.
-* **Comparison:** `operator<=>` is defaulted. With normalization, distinct representatives can compare as expected (e.g., `-180` vs `180` in signed ranges are distinct endpoints).
+* **Comparison:** `operator<=>` is defaulted. Normalization canonicalizes the maximum endpoint to the minimum endpoint.
 * **No implicit numeric conversion:** There’s **no `operator Type()`**. Use `as_degrees()`/`as_radians()` (or `*angle`) explicitly—this intentional friction avoids unit mistakes.
 
 ---

docs/trigonometry/angles.md

@@ -4,7 +4,7 @@
 > Namespace: `omath::angles`
 > All functions are `[[nodiscard]]` and `noexcept` where applicable.
 
-A small set of constexpr-friendly utilities for converting between degrees/radians, converting horizontal/vertical field of view, and wrapping angles into a closed interval.
+A small set of constexpr-friendly utilities for converting between degrees/radians, converting horizontal/vertical field of view, and wrapping angles into a half-open interval.
 
 ---
 
@@ -29,9 +29,9 @@ template<class Type>
 requires std::is_floating_point_v<Type>
 Type vertical_fov_to_horizontal(const Type& vertical_fov, const Type& aspect) noexcept;
 
-// Wrap angle into [min, max] (any arithmetic type)
+// Wrap angle into [min, max) (floating-point types)
 template<class Type>
-requires std::is_arithmetic_v<Type>
+requires std::is_floating_point_v<Type>
 Type wrap_angle(const Type& angle, const Type& min, const Type& max) noexcept;
 ```
 
@@ -66,10 +66,10 @@ Formulas (in radians):
 
 ### Wrapping angles (or any periodic value)
 
-Wrap any numeric `angle` into `[min, max]`:
+Wrap any floating-point `angle` into `[min, max)`:
 
 ```cpp
-// Wrap degrees into [0, 360]
+// Wrap degrees into [0, 360)
 float a = omath::angles::wrap_angle(   370.0f, 0.0f, 360.0f); // 10
 float b = omath::angles::wrap_angle(   -15.0f, 0.0f, 360.0f); // 345
 // Signed range [-180,180]
@@ -83,10 +83,10 @@ float c = omath::angles::wrap_angle(  200.0f, -180.0f, 180.0f); // -160
 * **Type requirements**
 
     * Converters & FOV helpers require **floating-point** `Type`.
-    * `wrap_angle` accepts any arithmetic `Type` (floats or integers).
+    * `wrap_angle` accepts floating-point types.
 * **Aspect ratio** must be **positive** and finite. For `aspect == 0` the FOV helpers are undefined.
 * **Units**: FOV functions accept/return **degrees** but compute internally in radians.
-* **Wrapping interval**: Behavior assumes `max > min`. The result lies in the **closed interval** `[min, max]` with modulo arithmetic; if you need half-open behavior (e.g., `[min,max)`), adjust your range or post-process endpoint cases.
+* **Wrapping interval**: Behavior assumes `max > min`. The result lies in the half-open interval `[min, max)`.
 * **constexpr**: Converters are `constexpr`; FOV helpers are runtime constexpr-compatible except for `std::atan/std::tan` constraints on some standard libraries.
 
 ---
@@ -103,5 +103,5 @@ float v = horizontal_fov_to_vertical(90.0f, 16.0f/9.0f);
 float h = vertical_fov_to_horizontal(v, 16.0f/9.0f);
 assert(std::abs(h - 90.0f) < 1e-5f);
 
-assert(wrap_angle(360.0f, 0.0f, 360.0f) == 0.0f || wrap_angle(360.0f, 0.0f, 360.0f) == 360.0f);
+assert(wrap_angle(360.0f, 0.0f, 360.0f) == 0.0f);
 ```

examples/example_hud/gui/main_window.cpp

@@ -161,6 +161,12 @@ namespace imgui_desktop::gui
                                   ImGuiColorEditFlags_NoInputs);
                 ImGui::ColorEdit4("Gradient right##lbl", reinterpret_cast<float*>(&m_label_gradient_right),
                                   ImGuiColorEditFlags_NoInputs);
+                if (m_animate_gradients)
+                {
+                    int direction = static_cast<int>(m_label_gradient_direction);
+                    if (ImGui::Combo("Direction##lbl", &direction, "Right to left\0Left to right\0"))
+                        m_label_gradient_direction = static_cast<omath::hud::GradientDirection>(direction);
+                }
             }
             ImGui::SliderFloat("Offset##lbl", &m_label_offset, 0.f, 15.f);
             ImGui::Checkbox("Right##lbl", &m_show_right_labels);
@@ -309,7 +315,8 @@ namespace imgui_desktop::gui
         const Paint centered_label_color =
                 m_gradient_label ? Paint{omath::hud::Gradient{m_label_gradient_left, m_label_gradient_right,
                                                               m_label_gradient_right, m_label_gradient_left,
-                                                              m_animate_gradients}}
+                                                              m_animate_gradients, 4.f, 0.7f,
+                                                              m_label_gradient_direction}}
                                  : Paint{omath::Color::from_rgba(255, 255, 255, 255)};
 
         auto outline_helper = [](const bool is_outline) -> Outlined

examples/example_hud/gui/main_window.hpp

@@ -68,6 +68,7 @@ namespace imgui_desktop::gui
         omath::Color m_label_gradient_right{1.f, 0.2f, 0.7f, 1.f};
         bool m_outlined = true;
         bool m_gradient_label = true;
+        omath::hud::GradientDirection m_label_gradient_direction = omath::hud::GradientDirection::RightToLeft;
         bool m_show_right_labels = true, m_show_left_labels = true;
         bool m_show_top_labels = true, m_show_bottom_labels = true;
         bool m_show_centered_top = true, m_show_centered_bottom = true;

include/omath/engines/cry_engine/traits/camera_trait.hpp

@@ -4,7 +4,6 @@
 
 #pragma once
 #include "omath/engines/cry_engine/formulas.hpp"
-#include "omath/internal/constexpr_math.hpp"
 #include "omath/projection/camera.hpp"
 namespace omath::cry_engine
 {
@@ -16,8 +15,8 @@ namespace omath::cry_engine
                                                        const Vector3<float>& look_at) noexcept
         {
             const auto direction = (look_at - cam_origin).normalized();
-            return {PitchAngle::from_radians(internal::asin(direction.z)),
-                    YawAngle::from_radians(-internal::atan2(direction.x, direction.y)), RollAngle::from_radians(0.f)};
+            return {PitchAngle::from_asin(direction.z), -YawAngle::from_atan2(direction.x, direction.y),
+                    RollAngle::from_radians(0.f)};
         }
 
         [[nodiscard("view matrix result should not be discarded")]]

include/omath/engines/frostbite_engine/traits/camera_trait.hpp

@@ -4,7 +4,6 @@
 
 #pragma once
 #include "omath/engines/frostbite_engine/formulas.hpp"
-#include "omath/internal/constexpr_math.hpp"
 #include "omath/projection/camera.hpp"
 
 namespace omath::frostbite_engine
@@ -18,8 +17,8 @@ namespace omath::frostbite_engine
         {
             const auto direction = (look_at - cam_origin).normalized();
 
-            return {PitchAngle::from_radians(-internal::asin(direction.y)),
-                    YawAngle::from_radians(internal::atan2(direction.x, direction.z)), RollAngle::from_radians(0.f)};
+            return {-PitchAngle::from_asin(direction.y), YawAngle::from_atan2(direction.x, direction.z),
+                    RollAngle::from_radians(0.f)};
         }
 
         [[nodiscard("view matrix result should not be discarded")]]

include/omath/engines/iw_engine/traits/camera_trait.hpp

@@ -4,7 +4,6 @@
 
 #pragma once
 #include "omath/engines/iw_engine/formulas.hpp"
-#include "omath/internal/constexpr_math.hpp"
 #include "omath/projection/camera.hpp"
 
 namespace omath::iw_engine
@@ -18,8 +17,8 @@ namespace omath::iw_engine
         {
             const auto direction = (look_at - cam_origin).normalized();
 
-            return {PitchAngle::from_radians(-internal::asin(direction.z)),
-                    YawAngle::from_radians(internal::atan2(direction.y, direction.x)), RollAngle::from_radians(0.f)};
+            return {-PitchAngle::from_asin(direction.z), YawAngle::from_atan2(direction.y, direction.x),
+                    RollAngle::from_radians(0.f)};
         }
 
         [[nodiscard("view matrix result should not be discarded")]]

include/omath/engines/opengl_engine/traits/camera_trait.hpp

@@ -4,7 +4,6 @@
 
 #pragma once
 #include "omath/engines/opengl_engine/formulas.hpp"
-#include "omath/internal/constexpr_math.hpp"
 #include "omath/projection/camera.hpp"
 
 namespace omath::opengl_engine
@@ -18,8 +17,8 @@ namespace omath::opengl_engine
         {
             const auto direction = (look_at - cam_origin).normalized();
 
-            return {PitchAngle::from_radians(internal::asin(direction.y)),
-                    YawAngle::from_radians(-internal::atan2(direction.x, -direction.z)), RollAngle::from_radians(0.f)};
+            return {PitchAngle::from_asin(direction.y), -YawAngle::from_atan2(direction.x, -direction.z),
+                    RollAngle::from_radians(0.f)};
         }
 
         [[nodiscard("view matrix result should not be discarded")]]

include/omath/engines/rage_engine/traits/camera_trait.hpp

@@ -4,7 +4,6 @@
 
 #pragma once
 #include "omath/engines/rage_engine/formulas.hpp"
-#include "omath/internal/constexpr_math.hpp"
 #include "omath/projection/camera.hpp"
 
 namespace omath::rage_engine
@@ -18,8 +17,8 @@ namespace omath::rage_engine
         {
             const auto direction = (look_at - cam_origin).normalized();
 
-            return {PitchAngle::from_radians(internal::asin(direction.z)),
-                    YawAngle::from_radians(-internal::atan2(direction.x, direction.y)), RollAngle::from_radians(0.f)};
+            return {PitchAngle::from_asin(direction.z), -YawAngle::from_atan2(direction.x, direction.y),
+                    RollAngle::from_radians(0.f)};
         }
 
         [[nodiscard("view matrix result should not be discarded")]]

include/omath/engines/source_engine/traits/camera_trait.hpp

@@ -4,7 +4,6 @@
 
 #pragma once
 #include "omath/engines/source_engine/formulas.hpp"
-#include "omath/internal/constexpr_math.hpp"
 #include "omath/projection/camera.hpp"
 
 namespace omath::source_engine
@@ -18,8 +17,8 @@ namespace omath::source_engine
         {
             const auto direction = (look_at - cam_origin).normalized();
 
-            return {PitchAngle::from_radians(-internal::asin(direction.z)),
-                    YawAngle::from_radians(internal::atan2(direction.y, direction.x)), RollAngle::from_radians(0.f)};
+            return {-PitchAngle::from_asin(direction.z),
+                    YawAngle::from_atan2(direction.y, direction.x), RollAngle::from_radians(0.f)};
         }
 
         [[nodiscard("view matrix result should not be discarded")]]

include/omath/engines/unity_engine/traits/camera_trait.hpp

@@ -4,7 +4,6 @@
 
 #pragma once
 #include "omath/engines/unity_engine/formulas.hpp"
-#include "omath/internal/constexpr_math.hpp"
 #include "omath/projection/camera.hpp"
 
 namespace omath::unity_engine
@@ -18,8 +17,8 @@ namespace omath::unity_engine
         {
             const auto direction = (look_at - cam_origin).normalized();
 
-            return {PitchAngle::from_radians(-internal::asin(direction.y)),
-                    YawAngle::from_radians(internal::atan2(direction.x, direction.z)), RollAngle::from_radians(0.f)};
+            return {-PitchAngle::from_asin(direction.y), YawAngle::from_atan2(direction.x, direction.z),
+                    RollAngle::from_radians(0.f)};
         }
 
         [[nodiscard("view matrix result should not be discarded")]]

include/omath/engines/unreal_engine/traits/camera_trait.hpp

@@ -4,7 +4,6 @@
 
 #pragma once
 #include "omath/engines/unreal_engine/formulas.hpp"
-#include "omath/internal/constexpr_math.hpp"
 #include "omath/projection/camera.hpp"
 
 namespace omath::unreal_engine
@@ -18,8 +17,8 @@ namespace omath::unreal_engine
         {
             const auto direction = (look_at - cam_origin).normalized();
 
-            return {PitchAngle::from_radians(internal::asin(direction.z)),
-                    YawAngle::from_radians(internal::atan2(direction.y, direction.x)), RollAngle::from_radians(0.f)};
+            return {PitchAngle::from_asin(direction.z), YawAngle::from_atan2(direction.y, direction.x),
+                    RollAngle::from_radians(0.f)};
         }
 
         [[nodiscard("view matrix result should not be discarded")]]

include/omath/hud/gradient.hpp

@@ -3,7 +3,13 @@
 
 namespace omath::hud
 {
-    struct Gradient
+    enum class GradientDirection
+    {
+        RightToLeft,
+        LeftToRight,
+    };
+
+    struct Gradient final
     {
         Color top_left;
         Color top_right;
@@ -12,5 +18,6 @@ namespace omath::hud
         bool animated = false;
         float animation_speed = 4.f;
         float animation_spread = 0.7f;
+        GradientDirection direction = GradientDirection::RightToLeft;
     };
 } // namespace omath::hud

include/omath/linear_algebra/mat.hpp

@@ -186,7 +186,14 @@ namespace omath
                 else if constexpr (StoreType == MatStoreType::COLUMN_MAJOR)
                     return cache_friendly_multiply_col_major(other);
             }
-            if constexpr (StoreType == MatStoreType::ROW_MAJOR)
+            if constexpr (!std::is_same_v<Type, float> && !std::is_same_v<Type, double>)
+            {
+                if constexpr (StoreType == MatStoreType::ROW_MAJOR)
+                    return cache_friendly_multiply_row_major(other);
+                else if constexpr (StoreType == MatStoreType::COLUMN_MAJOR)
+                    return cache_friendly_multiply_col_major(other);
+            }
+            else if constexpr (StoreType == MatStoreType::ROW_MAJOR)
                 return avx_multiply_row_major(other);
             else if constexpr (StoreType == MatStoreType::COLUMN_MAJOR)
                 return avx_multiply_col_major(other);
@@ -429,13 +436,22 @@ namespace omath
         cache_friendly_multiply_row_major(const Mat<Columns, OtherColumns, Type, MatStoreType::ROW_MAJOR>& other) const
         {
             Mat<Rows, OtherColumns, Type, MatStoreType::ROW_MAJOR> result;
+            const Type* left_data = m_data.data();
+            const Type* right_data = other.raw_array().data();
+            Type* result_data = result.raw_array().data();
+
             for (std::size_t row_index = 0; row_index < Rows; ++row_index)
+            {
+                const Type* left_row = left_data + row_index * Columns;
+                Type* result_row = result_data + row_index * OtherColumns;
                 for (std::size_t column_index = 0; column_index < Columns; ++column_index)
                 {
-                    const Type& current_number = at(row_index, column_index);
+                    const Type current_number = left_row[column_index];
+                    const Type* right_row = right_data + column_index * OtherColumns;
                     for (std::size_t other_column = 0; other_column < OtherColumns; ++other_column)
-                        result.at(row_index, other_column) += current_number * other.at(column_index, other_column);
+                        result_row[other_column] += current_number * right_row[other_column];
                 }
+            }
             return result;
         }
 
@@ -444,13 +460,22 @@ namespace omath
                 const Mat<Columns, OtherColumns, Type, MatStoreType::COLUMN_MAJOR>& other) const
         {
             Mat<Rows, OtherColumns, Type, MatStoreType::COLUMN_MAJOR> result;
+            const Type* left_data = m_data.data();
+            const Type* right_data = other.raw_array().data();
+            Type* result_data = result.raw_array().data();
+
             for (std::size_t other_column = 0; other_column < OtherColumns; ++other_column)
+            {
+                const Type* right_column = right_data + other_column * Columns;
+                Type* result_column = result_data + other_column * Rows;
                 for (std::size_t column_index = 0; column_index < Columns; ++column_index)
                 {
-                    const Type& current_number = other.at(column_index, other_column);
+                    const Type current_number = right_column[column_index];
+                    const Type* left_column = left_data + column_index * Rows;
                     for (std::size_t row_index = 0; row_index < Rows; ++row_index)
-                        result.at(row_index, other_column) += at(row_index, column_index) * current_number;
+                        result_column[row_index] += left_column[row_index] * current_number;
                 }
+            }
             return result;
         }
 #ifdef OMATH_USE_AVX2
@@ -466,56 +491,92 @@ namespace omath
 
             if constexpr (std::is_same_v<Type, float>)
             {
-                // ReSharper disable once CppTooWideScopeInitStatement
                 constexpr std::size_t vector_size = 8;
+                constexpr std::size_t block_size = vector_size * 4;
                 for (std::size_t j = 0; j < OtherColumns; ++j)
                 {
                     auto* c_col = reinterpret_cast<float*>(result_mat_data + j * Rows);
-                    for (std::size_t k = 0; k < Columns; ++k)
+                    std::size_t i = 0;
+                    for (; i + block_size <= Rows; i += block_size)
                     {
-                        const float bkj = reinterpret_cast<const float*>(other_mat_data)[k + j * Columns];
-                        const __m256 bkj_vec = _mm256_set1_ps(bkj);
-
-                        const auto* a_col_k = reinterpret_cast<const float*>(this_mat_data + k * Rows);
-
-                        std::size_t i = 0;
-                        for (; i + vector_size <= Rows; i += vector_size)
+                        __m256 cvec0 = _mm256_setzero_ps();
+                        __m256 cvec1 = _mm256_setzero_ps();
+                        __m256 cvec2 = _mm256_setzero_ps();
+                        __m256 cvec3 = _mm256_setzero_ps();
+                        for (std::size_t k = 0; k < Columns; ++k)
                         {
-                            __m256 cvec = _mm256_loadu_ps(c_col + i);
+                            const __m256 bkj_vec = _mm256_set1_ps(other_mat_data[k + j * Columns]);
+                            const auto* a_col_k = this_mat_data + k * Rows + i;
+                            cvec0 = _mm256_fmadd_ps(_mm256_loadu_ps(a_col_k), bkj_vec, cvec0);
+                            cvec1 = _mm256_fmadd_ps(_mm256_loadu_ps(a_col_k + vector_size), bkj_vec, cvec1);
+                            cvec2 = _mm256_fmadd_ps(_mm256_loadu_ps(a_col_k + vector_size * 2), bkj_vec, cvec2);
+                            cvec3 = _mm256_fmadd_ps(_mm256_loadu_ps(a_col_k + vector_size * 3), bkj_vec, cvec3);
+                        }
+                        _mm256_storeu_ps(c_col + i, cvec0);
+                        _mm256_storeu_ps(c_col + i + vector_size, cvec1);
+                        _mm256_storeu_ps(c_col + i + vector_size * 2, cvec2);
+                        _mm256_storeu_ps(c_col + i + vector_size * 3, cvec3);
+                    }
+                    for (; i + vector_size <= Rows; i += vector_size)
+                    {
+                        __m256 cvec = _mm256_setzero_ps();
+                        for (std::size_t k = 0; k < Columns; ++k)
+                        {
+                            const __m256 bkj_vec = _mm256_set1_ps(other_mat_data[k + j * Columns]);
+                            const auto* a_col_k = this_mat_data + k * Rows;
                             const __m256 a_vec = _mm256_loadu_ps(a_col_k + i);
                             cvec = _mm256_fmadd_ps(a_vec, bkj_vec, cvec);
-                            _mm256_storeu_ps(c_col + i, cvec);
                         }
-                        for (; i < Rows; ++i)
-                            c_col[i] += a_col_k[i] * bkj;
+                        _mm256_storeu_ps(c_col + i, cvec);
                     }
+                    for (; i < Rows; ++i)
+                        for (std::size_t k = 0; k < Columns; ++k)
+                            c_col[i] += this_mat_data[i + k * Rows] * other_mat_data[k + j * Columns];
                 }
             }
             else if (std::is_same_v<Type, double>)
-            { // double
-                // ReSharper disable once CppTooWideScopeInitStatement
+            {
                 constexpr std::size_t vector_size = 4;
+                constexpr std::size_t block_size = vector_size * 4;
                 for (std::size_t j = 0; j < OtherColumns; ++j)
                 {
                     auto* c_col = reinterpret_cast<double*>(result_mat_data + j * Rows);
-                    for (std::size_t k = 0; k < Columns; ++k)
+                    std::size_t i = 0;
+                    for (; i + block_size <= Rows; i += block_size)
                     {
-                        const double bkj = reinterpret_cast<const double*>(other_mat_data)[k + j * Columns];
-                        const __m256d bkj_vec = _mm256_set1_pd(bkj);
-
-                        const auto* a_col_k = reinterpret_cast<const double*>(this_mat_data + k * Rows);
-
-                        std::size_t i = 0;
-                        for (; i + vector_size <= Rows; i += vector_size)
+                        __m256d cvec0 = _mm256_setzero_pd();
+                        __m256d cvec1 = _mm256_setzero_pd();
+                        __m256d cvec2 = _mm256_setzero_pd();
+                        __m256d cvec3 = _mm256_setzero_pd();
+                        for (std::size_t k = 0; k < Columns; ++k)
                         {
-                            __m256d cvec = _mm256_loadu_pd(c_col + i);
+                            const __m256d bkj_vec = _mm256_set1_pd(other_mat_data[k + j * Columns]);
+                            const auto* a_col_k = this_mat_data + k * Rows + i;
+                            cvec0 = _mm256_fmadd_pd(_mm256_loadu_pd(a_col_k), bkj_vec, cvec0);
+                            cvec1 = _mm256_fmadd_pd(_mm256_loadu_pd(a_col_k + vector_size), bkj_vec, cvec1);
+                            cvec2 = _mm256_fmadd_pd(_mm256_loadu_pd(a_col_k + vector_size * 2), bkj_vec, cvec2);
+                            cvec3 = _mm256_fmadd_pd(_mm256_loadu_pd(a_col_k + vector_size * 3), bkj_vec, cvec3);
+                        }
+                        _mm256_storeu_pd(c_col + i, cvec0);
+                        _mm256_storeu_pd(c_col + i + vector_size, cvec1);
+                        _mm256_storeu_pd(c_col + i + vector_size * 2, cvec2);
+                        _mm256_storeu_pd(c_col + i + vector_size * 3, cvec3);
+                    }
+                    for (; i + vector_size <= Rows; i += vector_size)
+                    {
+                        __m256d cvec = _mm256_setzero_pd();
+                        for (std::size_t k = 0; k < Columns; ++k)
+                        {
+                            const __m256d bkj_vec = _mm256_set1_pd(other_mat_data[k + j * Columns]);
+                            const auto* a_col_k = this_mat_data + k * Rows;
                             const __m256d a_vec = _mm256_loadu_pd(a_col_k + i);
                             cvec = _mm256_fmadd_pd(a_vec, bkj_vec, cvec);
-                            _mm256_storeu_pd(c_col + i, cvec);
                         }
-                        for (; i < Rows; ++i)
-                            c_col[i] += a_col_k[i] * bkj;
+                        _mm256_storeu_pd(c_col + i, cvec);
                     }
+                    for (; i < Rows; ++i)
+                        for (std::size_t k = 0; k < Columns; ++k)
+                            c_col[i] += this_mat_data[i + k * Rows] * other_mat_data[k + j * Columns];
                 }
             }
             else
@@ -536,56 +597,92 @@ namespace omath
 
             if constexpr (std::is_same_v<Type, float>)
             {
-                // ReSharper disable once CppTooWideScopeInitStatement
                 constexpr std::size_t vector_size = 8;
+                constexpr std::size_t block_size = vector_size * 4;
                 for (std::size_t i = 0; i < Rows; ++i)
                 {
-                    Type* c_row = result_mat_data + i * OtherColumns;
-                    for (std::size_t k = 0; k < Columns; ++k)
+                    auto* c_row = reinterpret_cast<float*>(result_mat_data + i * OtherColumns);
+                    std::size_t j = 0;
+                    for (; j + block_size <= OtherColumns; j += block_size)
                     {
-                        const auto aik = static_cast<float>(this_mat_data[i * Columns + k]);
-                        const __m256 aik_vec = _mm256_set1_ps(aik);
-                        const auto* b_row = reinterpret_cast<const float*>(other_mat_data + k * OtherColumns);
-
-                        std::size_t j = 0;
-                        for (; j + vector_size <= OtherColumns; j += vector_size)
+                        __m256 cvec0 = _mm256_setzero_ps();
+                        __m256 cvec1 = _mm256_setzero_ps();
+                        __m256 cvec2 = _mm256_setzero_ps();
+                        __m256 cvec3 = _mm256_setzero_ps();
+                        for (std::size_t k = 0; k < Columns; ++k)
                         {
-                            __m256 cvec = _mm256_loadu_ps(c_row + j);
+                            const __m256 aik_vec = _mm256_set1_ps(this_mat_data[i * Columns + k]);
+                            const auto* b_row = other_mat_data + k * OtherColumns + j;
+                            cvec0 = _mm256_fmadd_ps(_mm256_loadu_ps(b_row), aik_vec, cvec0);
+                            cvec1 = _mm256_fmadd_ps(_mm256_loadu_ps(b_row + vector_size), aik_vec, cvec1);
+                            cvec2 = _mm256_fmadd_ps(_mm256_loadu_ps(b_row + vector_size * 2), aik_vec, cvec2);
+                            cvec3 = _mm256_fmadd_ps(_mm256_loadu_ps(b_row + vector_size * 3), aik_vec, cvec3);
+                        }
+                        _mm256_storeu_ps(c_row + j, cvec0);
+                        _mm256_storeu_ps(c_row + j + vector_size, cvec1);
+                        _mm256_storeu_ps(c_row + j + vector_size * 2, cvec2);
+                        _mm256_storeu_ps(c_row + j + vector_size * 3, cvec3);
+                    }
+                    for (; j + vector_size <= OtherColumns; j += vector_size)
+                    {
+                        __m256 cvec = _mm256_setzero_ps();
+                        for (std::size_t k = 0; k < Columns; ++k)
+                        {
+                            const __m256 aik_vec = _mm256_set1_ps(this_mat_data[i * Columns + k]);
+                            const auto* b_row = other_mat_data + k * OtherColumns;
                             const __m256 b_vec = _mm256_loadu_ps(b_row + j);
                             cvec = _mm256_fmadd_ps(b_vec, aik_vec, cvec);
-
-                            _mm256_storeu_ps(c_row + j, cvec);
                         }
-                        for (; j < OtherColumns; ++j)
-                            c_row[j] += aik * b_row[j];
+                        _mm256_storeu_ps(c_row + j, cvec);
                     }
+                    for (; j < OtherColumns; ++j)
+                        for (std::size_t k = 0; k < Columns; ++k)
+                            c_row[j] += this_mat_data[i * Columns + k] * other_mat_data[k * OtherColumns + j];
                 }
             }
             else if (std::is_same_v<Type, double>)
-            { // double
-                // ReSharper disable once CppTooWideScopeInitStatement
+            {
                 constexpr std::size_t vector_size = 4;
+                constexpr std::size_t block_size = vector_size * 4;
                 for (std::size_t i = 0; i < Rows; ++i)
                 {
-                    Type* c_row = result_mat_data + i * OtherColumns;
-                    for (std::size_t k = 0; k < Columns; ++k)
+                    auto* c_row = reinterpret_cast<double*>(result_mat_data + i * OtherColumns);
+                    std::size_t j = 0;
+                    for (; j + block_size <= OtherColumns; j += block_size)
                     {
-                        const auto aik = static_cast<double>(this_mat_data[i * Columns + k]);
-                        const __m256d aik_vec = _mm256_set1_pd(aik);
-                        const auto* b_row = reinterpret_cast<const double*>(other_mat_data + k * OtherColumns);
-
-                        std::size_t j = 0;
-                        for (; j + vector_size <= OtherColumns; j += vector_size)
+                        __m256d cvec0 = _mm256_setzero_pd();
+                        __m256d cvec1 = _mm256_setzero_pd();
+                        __m256d cvec2 = _mm256_setzero_pd();
+                        __m256d cvec3 = _mm256_setzero_pd();
+                        for (std::size_t k = 0; k < Columns; ++k)
                         {
-                            __m256d cvec = _mm256_loadu_pd(c_row + j);
+                            const __m256d aik_vec = _mm256_set1_pd(this_mat_data[i * Columns + k]);
+                            const auto* b_row = other_mat_data + k * OtherColumns + j;
+                            cvec0 = _mm256_fmadd_pd(_mm256_loadu_pd(b_row), aik_vec, cvec0);
+                            cvec1 = _mm256_fmadd_pd(_mm256_loadu_pd(b_row + vector_size), aik_vec, cvec1);
+                            cvec2 = _mm256_fmadd_pd(_mm256_loadu_pd(b_row + vector_size * 2), aik_vec, cvec2);
+                            cvec3 = _mm256_fmadd_pd(_mm256_loadu_pd(b_row + vector_size * 3), aik_vec, cvec3);
+                        }
+                        _mm256_storeu_pd(c_row + j, cvec0);
+                        _mm256_storeu_pd(c_row + j + vector_size, cvec1);
+                        _mm256_storeu_pd(c_row + j + vector_size * 2, cvec2);
+                        _mm256_storeu_pd(c_row + j + vector_size * 3, cvec3);
+                    }
+                    for (; j + vector_size <= OtherColumns; j += vector_size)
+                    {
+                        __m256d cvec = _mm256_setzero_pd();
+                        for (std::size_t k = 0; k < Columns; ++k)
+                        {
+                            const __m256d aik_vec = _mm256_set1_pd(this_mat_data[i * Columns + k]);
+                            const auto* b_row = other_mat_data + k * OtherColumns;
                             const __m256d b_vec = _mm256_loadu_pd(b_row + j);
                             cvec = _mm256_fmadd_pd(b_vec, aik_vec, cvec);
-
-                            _mm256_storeu_pd(c_row + j, cvec);
                         }
-                        for (; j < OtherColumns; ++j)
-                            c_row[j] += aik * b_row[j];
+                        _mm256_storeu_pd(c_row + j, cvec);
                     }
+                    for (; j < OtherColumns; ++j)
+                        for (std::size_t k = 0; k < Columns; ++k)
+                            c_row[j] += this_mat_data[i * Columns + k] * other_mat_data[k * OtherColumns + j];
                 }
             }
             else

include/omath/trigonometry/angle.hpp

@@ -6,7 +6,9 @@
 #include "omath/internal/constexpr_math.hpp"
 #include "omath/trigonometry/angles.hpp"
 #include <algorithm>
+#include <compare>
 #include <format>
+#include <type_traits>
 #include <utility>
 
 namespace omath
@@ -18,7 +20,7 @@ namespace omath
     };
 
     template<class Type = float, Type min = Type(0), Type max = Type(360), AngleFlags flags = AngleFlags::Normalized>
-    requires std::is_arithmetic_v<Type>
+    requires std::is_floating_point_v<Type>
     class Angle
     {
         Type m_angle;
@@ -43,7 +45,7 @@ namespace omath
         {
             return Angle{degrees};
         }
-        constexpr Angle() noexcept: m_angle(0)
+        constexpr Angle() noexcept: Angle(Type{0})
         {
         }
         [[nodiscard]]
@@ -52,6 +54,30 @@ namespace omath
             return Angle{angles::radians_to_degrees<Type>(degrees)};
         }
 
+        [[nodiscard]]
+        constexpr static Angle from_asin(const Type& value) noexcept
+        {
+            return from_radians(internal::asin(value));
+        }
+
+        [[nodiscard]]
+        constexpr static Angle from_acos(const Type& value) noexcept
+        {
+            return from_radians(internal::acos(value));
+        }
+
+        [[nodiscard]]
+        constexpr static Angle from_atan(const Type& value) noexcept
+        {
+            return from_radians(internal::atan(value));
+        }
+
+        [[nodiscard]]
+        constexpr static Angle from_atan2(const Type& y, const Type& x) noexcept
+        {
+            return from_radians(internal::atan2(y, x));
+        }
+
         [[nodiscard]]
         constexpr const Type& operator*() const noexcept
         {
@@ -88,12 +114,6 @@ namespace omath
             return internal::tan(as_radians());
         }
 
-        [[nodiscard]]
-        constexpr Type atan() const noexcept
-        {
-            return internal::atan(as_radians());
-        }
-
         [[nodiscard]]
         constexpr Type cot() const noexcept
         {
@@ -121,7 +141,8 @@ namespace omath
 
         constexpr Angle& operator-=(const Angle& other) noexcept
         {
-            return operator+=(-other);
+            *this = Angle{m_angle - other.m_angle};
+            return *this;
         }
 
         [[nodiscard]]
@@ -142,7 +163,7 @@ namespace omath
         [[nodiscard]]
         constexpr Angle operator-(const Angle& other) const noexcept
         {
-            return operator+(-other);
+            return Angle{m_angle - other.m_angle};
         }
 
         [[nodiscard]]
@@ -172,7 +193,6 @@ struct std::formatter<omath::Angle<T, MinV, MaxV, F>, char> final // NOLINT(*-dc
         return std::format_to(ctx.out(), "{}deg", a.as_degrees());
     }
 };
-
 // wchar_t formatter
 template<class T, T MinV, T MaxV, omath::AngleFlags F>
 struct std::formatter<omath::Angle<T, MinV, MaxV, F>, wchar_t> final // NOLINT(*-dcl58-cpp)
@@ -193,24 +213,3 @@ struct std::formatter<omath::Angle<T, MinV, MaxV, F>, wchar_t> final // NOLINT(*
         return std::format_to(ctx.out(), L"{}deg", a.as_degrees());
     }
 };
-
-// wchar_t formatter
-template<class T, T MinV, T MaxV, omath::AngleFlags F>
-struct std::formatter<omath::Angle<T, MinV, MaxV, F>, char8_t> final // NOLINT(*-dcl58-cpp)
-{
-    using AngleT = omath::Angle<T, MinV, MaxV, F>;
-
-    [[nodiscard]]
-    static constexpr auto parse(std::wformat_parse_context& ctx)
-    {
-        return ctx.begin();
-    }
-
-    template<class FormatContext>
-    [[nodiscard]]
-    auto format(const AngleT& a, FormatContext& ctx) const
-    {
-        static_assert(std::is_same_v<typename FormatContext::char_type, char8_t>);
-        return std::format_to(ctx.out(), u8"{}deg", a.as_degrees());
-    }
-};

include/omath/trigonometry/angles.hpp

@@ -48,10 +48,10 @@ namespace omath::angles
     }
 
     template<class Type>
-    requires std::is_arithmetic_v<Type>
+    requires std::is_floating_point_v<Type>
     [[nodiscard]] constexpr Type wrap_angle(const Type& angle, const Type& min, const Type& max) noexcept
     {
-        if (angle <= max && angle >= min)
+        if (angle < max && angle >= min)
             return angle;
 
         const Type range = max - min;

include/omath/utility/pattern_scan.hpp

@@ -54,9 +54,6 @@ namespace omath
                 return {value, Length - 1};
             }
         };
-        template<std::size_t N>
-        ConstevalPattern(const char (&)[N]) -> ConstevalPattern<N>;
-
         [[nodiscard]]
         static std::span<std::byte>::iterator scan_for_pattern(const std::span<std::byte>& range,
                                                                const std::string_view& pattern);

scripts/coverage-llvm.sh

@@ -177,9 +177,7 @@ if command -v genhtml >/dev/null 2>&1; then
         --title "Omath Coverage Report" \
         --show-details \
         --legend \
-        --demangle-cpp \
         --num-spaces 4 \
-        --sort \
         --function-coverage \
         --branch-coverage
     

source/hud/renderer_realizations/imgui_renderer.cpp

@@ -103,6 +103,7 @@ namespace omath::hud
         if (gradient.animated)
         {
             const float animation_offset = static_cast<float>(ImGui::GetTime()) * gradient.animation_speed;
+            const float direction = gradient.direction == GradientDirection::RightToLeft ? 1.f : -1.f;
             float cursor_x = position.x;
             const char* glyph_start = text.data();
             const char* const text_end = text.data() + text.size();
@@ -117,7 +118,9 @@ namespace omath::hud
 
                 const char* const glyph_end = glyph_start + glyph_size;
                 const float blend =
-                        (std::sin(animation_offset + static_cast<float>(glyph_index) * gradient.animation_spread) + 1.f)
+                        (std::sin(animation_offset + direction * static_cast<float>(glyph_index)
+                                                             * gradient.animation_spread)
+                         + 1.f)
                         * 0.5f;
                 const auto color = gradient.top_left.value() * (1.f - blend) + gradient.top_right.value() * blend;
                 draw_list->AddText({cursor_x, position.y}, Color{color}.to_im_color(), glyph_start, glyph_end);

tests/general/unit_test_angle.cpp

@@ -14,9 +14,14 @@ namespace
 
     // Handy aliases (defaults: Type=float, [0,360], Normalized)
     using Deg = Angle<float, static_cast<float>(0), static_cast<float>(360), AngleFlags::Normalized>;
+    using Fov = Angle<float, static_cast<float>(0), static_cast<float>(180), AngleFlags::Clamped>;
+    using Offset = Angle<float, static_cast<float>(10), static_cast<float>(20), AngleFlags::Clamped>;
     using Pitch = Angle<float, static_cast<float>(-90), static_cast<float>(90), AngleFlags::Clamped>;
     using Turn = Angle<float, static_cast<float>(-180), static_cast<float>(180), AngleFlags::Normalized>;
 
+    template<class Type>
+    concept SupportedAngleType = requires { typename Angle<Type>; };
+
     constexpr float k_eps = 1e-5f;
 
     constexpr bool close_to(const float actual, const float expected, const float epsilon)
@@ -36,6 +41,12 @@ TEST(UnitTestAngle, DefaultConstructor_IsZeroDegrees)
     EXPECT_FLOAT_EQ(a.as_degrees(), 0.0f);
 }
 
+TEST(UnitTestAngle, DefaultConstructor_AppliesRangePolicy)
+{
+    constexpr Offset a;
+    EXPECT_FLOAT_EQ(a.as_degrees(), 10.0f);
+}
+
 TEST(UnitTestAngle, FromDegrees_Normalized_WrapsAboveMax)
 {
     const Deg a = Deg::from_degrees(370.0f);
@@ -66,6 +77,14 @@ TEST(UnitTestAngle, FromRadians_And_AsRadians)
     EXPECT_NEAR(b.as_radians(), std::numbers::pi_v<float>, 1e-6f);
 }
 
+TEST(UnitTestAngle, FromInverseTrigonometricFunctions)
+{
+    EXPECT_NEAR(Pitch::from_asin(0.5f).as_degrees(), 30.0f, k_eps);
+    EXPECT_NEAR(Pitch::from_acos(0.5f).as_degrees(), 60.0f, k_eps);
+    EXPECT_NEAR(Pitch::from_atan(1.0f).as_degrees(), 45.0f, k_eps);
+    EXPECT_NEAR(Turn::from_atan2(-1.0f, -1.0f).as_degrees(), -135.0f, k_eps);
+}
+
 // ---------- Unary minus & deref ----------
 
 TEST(UnitTestAngle, UnaryMinus_Normalized)
@@ -101,17 +120,6 @@ TEST(UnitTestAngle, SinCosTanCot_BasicCases)
     EXPECT_NEAR(a90.cos(), 0.0f, 1e-4f);
 }
 
-TEST(UnitTestAngle, Atan_IsAtanOfRadians)
-{
-    // atan(as_radians). For 0° -> atan(0)=0.
-    const Deg a0 = Deg::from_degrees(0.0f);
-    EXPECT_NEAR(a0.atan(), 0.0f, k_eps);
-
-    const Deg a45 = Deg::from_degrees(45.0f);
-    // atan(pi/4) ≈ 0.665773...
-    EXPECT_NEAR(a45.atan(), 0.66577375f, 1e-6f);
-}
-
 // ---------- Compound arithmetic ----------
 
 TEST(UnitTestAngle, PlusEquals_Normalized_Wraps)
@@ -142,6 +150,16 @@ TEST(UnitTestAngle, MinusEquals_Clamped_Clamps)
     EXPECT_FLOAT_EQ(p.as_degrees(), -90.0f);
 }
 
+TEST(UnitTestAngle, Subtraction_ClampedNonSymmetricRange)
+{
+    Fov compound = Fov::from_degrees(90.0f);
+    compound -= Fov::from_degrees(10.0f);
+    EXPECT_FLOAT_EQ(compound.as_degrees(), 80.0f);
+
+    const Fov binary = Fov::from_degrees(90.0f) - Fov::from_degrees(10.0f);
+    EXPECT_FLOAT_EQ(binary.as_degrees(), 80.0f);
+}
+
 // ---------- Alternative ranges ----------
 
 TEST(UnitTestAngle, NormalizedRange_Neg180To180)
@@ -205,5 +223,12 @@ static_assert(close_to(Pitch::from_degrees(45.0f).tan(), 1.0f, 1e-4f),
               "Tan should be constexpr with embedded constexpr math");
 static_assert(close_to(Pitch::from_degrees(45.0f).cot(), 1.0f, 1e-4f),
               "Cot should be constexpr with embedded constexpr math");
-static_assert(close_to(Pitch::from_degrees(45.0f).atan(), 0.66577375f, 1e-6f),
-              "Atan should be constexpr with embedded constexpr math");
+static_assert(close_to(Pitch::from_asin(0.5f).as_degrees(), 30.0f, k_eps),
+              "From asin should be constexpr with embedded constexpr math");
+static_assert(close_to(Pitch::from_acos(0.5f).as_degrees(), 60.0f, k_eps),
+              "From acos should be constexpr with embedded constexpr math");
+static_assert(close_to(Pitch::from_atan(1.0f).as_degrees(), 45.0f, k_eps),
+              "From atan should be constexpr with embedded constexpr math");
+static_assert(close_to(Turn::from_atan2(-1.0f, -1.0f).as_degrees(), -135.0f, k_eps),
+              "From atan2 should be constexpr with embedded constexpr math");
+static_assert(!SupportedAngleType<int>, "Angle should only accept floating-point types");

tests/general/unit_test_angles.cpp

@@ -46,4 +46,11 @@ TEST(unit_test_angles, wrap_angle_negative_range)
     const float wrapped = omath::angles::wrap_angle(-90.f, 0.f, 360.f);
 
     EXPECT_NEAR(wrapped, 270.f, 0.01f);
-}
+}
+
+TEST(unit_test_angles, wrap_angle_maximum_maps_to_minimum)
+{
+    const float wrapped = omath::angles::wrap_angle(360.f, 0.f, 360.f);
+
+    EXPECT_FLOAT_EQ(wrapped, 0.f);
+}

tests/general/unit_test_mat.cpp

@@ -16,6 +16,21 @@ namespace
         const float diff = actual - expected;
         return (diff < 0.0f ? -diff : diff) <= epsilon;
     }
+
+    template<size_t Rows, size_t Columns, size_t OtherColumns, class Type, MatStoreType StoreType>
+    void expect_multiplication_matches_scalar_reference(const Mat<Rows, Columns, Type, StoreType>& left,
+                                                        const Mat<Columns, OtherColumns, Type, StoreType>& right)
+    {
+        const auto result = left * right;
+        for (size_t row = 0; row < Rows; ++row)
+            for (size_t column = 0; column < OtherColumns; ++column)
+            {
+                Type expected{};
+                for (size_t shared_index = 0; shared_index < Columns; ++shared_index)
+                    expected += left.at(row, shared_index) * right.at(shared_index, column);
+                EXPECT_EQ(result.at(row, column), expected);
+            }
+    }
 } // namespace
 
 class UnitTestMat : public ::testing::Test
@@ -92,6 +107,45 @@ TEST_F(UnitTestMat, Operator_Multiplication_Matrix)
     EXPECT_FLOAT_EQ(m3.at(1, 1), 22.0f);
 }
 
+TEST(UnitTestMatStandalone, Operator_Multiplication_RowMajorSimdAndTail)
+{
+    Mat<3, 5, float, MatStoreType::ROW_MAJOR> left;
+    Mat<5, 33, float, MatStoreType::ROW_MAJOR> right;
+    for (size_t row = 0; row < left.row_count(); ++row)
+        for (size_t column = 0; column < left.columns_count(); ++column)
+            left.at(row, column) = static_cast<float>(row * 3 + column + 1);
+    for (size_t row = 0; row < right.row_count(); ++row)
+        for (size_t column = 0; column < right.columns_count(); ++column)
+            right.at(row, column) = static_cast<float>((row + 1) * (column % 5 + 1));
+
+    expect_multiplication_matches_scalar_reference(left, right);
+}
+
+TEST(UnitTestMatStandalone, Operator_Multiplication_ColumnMajorSimdAndTail)
+{
+    Mat<17, 5, double, MatStoreType::COLUMN_MAJOR> left;
+    Mat<5, 3, double, MatStoreType::COLUMN_MAJOR> right;
+    for (size_t row = 0; row < left.row_count(); ++row)
+        for (size_t column = 0; column < left.columns_count(); ++column)
+            left.at(row, column) = static_cast<double>(row * 3 + column + 1);
+    for (size_t row = 0; row < right.row_count(); ++row)
+        for (size_t column = 0; column < right.columns_count(); ++column)
+            right.at(row, column) = static_cast<double>((row + 1) * (column + 1));
+
+    expect_multiplication_matches_scalar_reference(left, right);
+}
+
+TEST(UnitTestMatStandalone, Operator_Multiplication_IntegerFallsBackFromAvx)
+{
+    constexpr Mat<2, 3, int> left{{1, 2, 3}, {4, 5, 6}};
+    constexpr Mat<3, 2, int> right{{7, 8}, {9, 10}, {11, 12}};
+    constexpr auto result = left * right;
+    static_assert(result.at(0, 0) == 58);
+    static_assert(result.at(1, 1) == 154);
+
+    expect_multiplication_matches_scalar_reference(left, right);
+}
+
 TEST_F(UnitTestMat, Operator_Multiplication_Scalar)
 {
     Mat<2, 2> m3 = m2 * 2.0f;

vcpkg-configuration.json

@@ -1,7 +1,7 @@
 {
   "default-registry": {
     "kind": "git",
-    "baseline": "b1b19307e2d2ec1eefbdb7ea069de7d4bcd31f01",
+    "baseline": "0878b5224d4a4968940ee296a2e7fae2d3b62983",
     "repository": "https://github.com/microsoft/vcpkg"
   },
   "registries": [

vcpkg.json

@@ -21,7 +21,11 @@
       "dependencies": [
         {
           "name": "omath",
-          "features": ["imgui", "lua", "hooking"]
+          "features": [
+            "imgui",
+            "lua",
+            "hooking"
+          ]
         }
       ]
     },
@@ -50,16 +54,26 @@
         "opengl",
         {
           "name": "omath",
-          "features": ["hooking"],
+          "features": [
+            "hooking"
+          ],
           "platform": "windows & !arm & !uwp"
         },
         {
           "name": "imgui",
-          "features": ["glfw-binding", "opengl3-binding"]
+          "features": [
+            "glfw-binding",
+            "opengl3-binding"
+          ]
         },
         {
           "name": "imgui",
-          "features": ["dx9-binding", "dx11-binding", "dx12-binding", "win32-binding"],
+          "features": [
+            "dx9-binding",
+            "dx11-binding",
+            "dx12-binding",
+            "win32-binding"
+          ],
           "platform": "windows & !arm & !uwp"
         }
       ]