Fix #580 by using a fixed-point implementation for unit conversions using integer representations

example/measurement.cpp

@@ -70,15 +70,37 @@ class measurement {
   [[nodiscard]] friend constexpr measurement operator+(const measurement& lhs, const measurement& rhs)
   {
     using namespace std;
-    return measurement(lhs.value() + rhs.value(), hypot(lhs.uncertainty(), rhs.uncertainty()));
+    return measurement{std::in_place, lhs.value() + rhs.value(), hypot(lhs.uncertainty(), rhs.uncertainty())};
   }
 
   [[nodiscard]] friend constexpr measurement operator-(const measurement& lhs, const measurement& rhs)
   {
     using namespace std;
-    return measurement(lhs.value() - rhs.value(), hypot(lhs.uncertainty(), rhs.uncertainty()));
+    return measurement{std::in_place, lhs.value() - rhs.value(), hypot(lhs.uncertainty(), rhs.uncertainty())};
   }
 
+  template<typename To, mp_units::Magnitude M>
+  [[nodiscard]] constexpr measurement<To> scale(std::type_identity<measurement<To>>, M scaling_factor) const
+  {
+    constexpr std::type_identity<To> to_value_type;
+    return measurement<To>{
+      std::in_place,
+      mp_units::scale(to_value_type, scaling_factor, value()),
+      mp_units::scale(to_value_type, scaling_factor, value()),
+    };
+  }
+
+  template<mp_units::Magnitude M>
+  [[nodiscard]] constexpr auto scale(M scaling_factor) const
+  {
+    return measurement{
+      std::in_place,
+      mp_units::scale(scaling_factor, value()),
+      mp_units::scale(scaling_factor, value()),
+    };
+  }
+
+
   [[nodiscard]] friend constexpr measurement operator*(const measurement& lhs, const measurement& rhs)
   {
     const auto val = lhs.value() * rhs.value();
@@ -127,15 +149,23 @@ class measurement {
 private:
   value_type value_{};
   value_type uncertainty_{};
+
+  // NOLINTNEXTLINE(bugprone-easily-swappable-parameters)
+  constexpr measurement(std::in_place_t, value_type val, value_type err) :
+      value_(std::move(val)), uncertainty_(std::move(err))
+  {
+  }
 };
 
 }  // namespace
 
+
 template<typename T>
 constexpr bool mp_units::is_scalar<measurement<T>> = true;
 template<typename T>
 constexpr bool mp_units::is_vector<measurement<T>> = true;
 
+static_assert(mp_units::RepresentationOf<measurement<int>, mp_units::quantity_character::scalar>);
 static_assert(mp_units::RepresentationOf<measurement<double>, mp_units::quantity_character::scalar>);
 
 namespace {

src/core/CMakeLists.txt

@@ -33,12 +33,14 @@ add_mp_units_module(
     core mp-units-core
     HEADERS include/mp-units/bits/constexpr_math.h
             include/mp-units/bits/core_gmf.h
+            include/mp-units/bits/fixed_point.h
             include/mp-units/bits/get_associated_quantity.h
             include/mp-units/bits/hacks.h
             include/mp-units/bits/math_concepts.h
             include/mp-units/bits/module_macros.h
             include/mp-units/bits/quantity_spec_hierarchy.h
             include/mp-units/bits/ratio.h
+            include/mp-units/bits/scaling.h
             include/mp-units/bits/sudo_cast.h
             include/mp-units/bits/text_tools.h
             include/mp-units/bits/type_list.h

src/core/include/mp-units/bits/core_gmf.h

@@ -30,6 +30,7 @@
 
 #ifndef MP_UNITS_IMPORT_STD
 #include <array>
+#include <bit>
 #include <compare>
 #include <concepts>
 #include <cstddef>