What is a safe, cross-platform way to perform range checks before casting to a smaller numeric type?

Here is the closest duplicate I could find.

Despite the tags, the question seems to be about C, and the usable answer references the C99 spec.

What is the correct way to handle this check in C++98, without using Boost or other libraries?

You can copy the code from gsl::narrow() and tweak it slightly, turning it into can_narrow() returning a bool rather than throwing:

// narrow_cast(): a searchable way to do narrowing casts of values
template<class T, class U>
inline constexpr T narrow_cast(U u) noexcept
{ return static_cast<T>(u); }

namespace details
{
    template<class T, class U>
    struct is_same_signedness : public std::integral_constant<bool, std::is_signed<T>::value == std::is_signed<U>::value>
    {};
}

template<class T, class U>
inline bool can_narrow(U u)
{
    T t = narrow_cast<T>(u);
    if (static_cast<U>(t) != u)
        return false;
    if (!details::is_same_signedness<T, U>::value && ((t < T{}) != (u < U{})))
        return false;
    return true;
}