v15

• CMake package is now ARCH_INDEPENDENT, as it should be when it's
  header only. Thank you @PatrickKa for submitting the fix.

• Much reworked strong::range, which is now compatible with
  std::ranges. This was primarily sparked by a discussion with
  Harald Achitz. Thank you for bringing up the deficiencies.

v14

• Added several missing [[nodiscard]]

• .begin() and .end() are now constexpr for strong::range types. Thanks Harald Achitz for reporting the missing specifiers.

• Added .size() member to strong::range types, if the underlying type has a .size() member. Thanks Harald Achitz for the suggestion.

v13

• Added static_assert() messages explaining why a modifier cannot be
  used with an underlying type. Thank you @SilmaliTech for suggesting
  the improvement.

• Added modifier invocable for types with operator(). Thank you
  @DNKpp for the suggestion and the draft implementation.

v12

• Fixed a bug where fmt formatter function was not const, as it must be
  per the documentation. Thank you Tim Blechmann for submitting the
  fix.

• Strong types can be weakly_ordered, strongly_ordered and
  partially_ordered, using operator <=> in C++20 and later. Huge thanks
  to Björn Schäpers and Tim Blechmann for the suggestions, discussions
  and infinite patience.

v11

• A strong type can now be used as an NTTP (Non Type Template Parameter)

  using type = strong::type<int, struct int_>;
  
  template <type t>
  struct S {
      constexpr operator int() const { return value_of(t);}
  };
  
  int main()
  {
      S<type{3}> obj;
      return obj;
  }

  Thank you Toni Neubert for the implementation.

v10

• Corrected the test for, and use of, concepts when specializing
  std::numeric_limits<> for strong::arithmetic types. Thank you
  @HazardyKnusperkeks for reporting.

v9

• Lowered the minimum required CMake version to 3.7, unless unit-tests
  are built.

• A type predicate strong::type_is_v<type, modifier> has been introduced.
  It can be used in compile time tests for generic code to ensure that
  a strong type has the necessary modifiers. Example:

  using type = strong::type<int, struct type_,
                            strong::ordered,
                            strong::ordered_with<int, long>>;
  static_assert(strong::type_is_v<type, strong::ordered>);
  static_assert(strong::type_is_v<type, strong::ordered_with<long>>);

  Note that for variadic modifiers like strong::ordered_with<>, the
  predicate ensures that all types listed in the test are supported,
  so in the example above, the test type_is_v<type, ordered_with<long>>
  is true because 'long' is included in the type definition using
  ordered_with<int, long>.

  There's also a type version, to use as
  strong::is_type<type, modifier>::value;

• std::numeric_limits<T> is specialized for types using the
  strong::arithmetic modifier.

• Added missing const&& overloads for value_of()

v8

• Split the header file into one file for each modifier. This can
  in some cases speed up compilation times considerably. The header
  file <strong_type/strong_type.hpp> still includes everything, so
  no changes to existing code is necessary.

• Added new modifier scalable_with<Ts...>. Allows multiplication and
  division with each of the types Ts. If the underlying type can be
  divided, the result will be the first of the type in Ts.

• difference types are conditionally ordered, if the underlying type
  supports it. Previously they were always ordered, but e.g. a 2D
  vector on a plane is a useful difference type, but it is usually
  not ordered. This change allows the vector in 2D space to be used
  as a difference type.

• Add CMake alias target strong_type::strong_type

• Use the CMake option -DSTRONG_TYPE_UNIT_TEST=yes to build unit tests

v7

• Added missing #include for <fmt/ostream.h>. Thank you Björn Schäpers

• Fixed MSVC check (typo) for constexpr. Thank you Sergei Soloviev

• Builds cleanly with -Wconversion

• Strengthened the unit test suite and CI build structure

v6

• Added missing strong::is_ostreamable<> predicate needed for fmtlib 9. Thank you Tim Blechman @timblechmann for submitting the fix.