How to automatically add literal definitions, based on a single user-defined literal?

Question

C++11 offers user-defined literals. I've just started to play around with them, which made me wonder whether it would be possible to automatically add all SI multipliers to a single literal I define?

For example, if I define

Length operator "" _m(long double m) {
    return Length(m); // Length in meters
}

where Length is a subclass of some Units base class, I would like to have a mechanism to automatically add (in the same spirit as boost operators) SI multipliers for all literals that return a Length:

// these are added automatically when defining the literal "_m": 
                                         // Length in:
Length operator "" _Ym(long double Ym);  // Yottameters
Length operator "" _Zm(long double Zm);  // Zetameters
...                                      // ...
...                                      // ...
Length operator "" _km(long double km);  // kilometers
Length operator "" _mm(long double mm);  // millimeters
...                                      // ...       
...                                      // ...
Length operator "" _zm(long double zm);  // zeptometers
Length operator "" _ym(long double ym);  // yoctometers

As far as I could see, aside from perhaps some macro magic, there is no way to do this automatically since all user-defined literals need an explicit definition.

..or am I overlooking something?

Answer 1

I do not think there is a way to do exactly what you are asking for without "bizarre macros". This is as far as I could get:

template<typename T, T (*op)(long double)>
struct SI
{
    // ...
    constexpr static T micro = op (.000001);
    constexpr static T milli = op (.001);
    constexpr static T kilo = op (1000);
    constexpr static T mega = op (1000000);
    // ...
};

struct Length
{
    constexpr Length(long double d) : _d(d) { }
    constexpr operator long double() { return _d; }
    long double _d;
};

constexpr Length operator "" _m(long double m) {
    return Length(m);
}

typedef SI<Length, ::operator "" _m> SI_Length;

int main()
{
    constexpr Length l = 3 * SI_Length::kilo;
    static_assert(l == 3000, "error");
}

If bizarre macros are allowed, then something like the following should do the job:

#define DEFINE_SI_MULTIPLIERS(T, unit) \
    constexpr T operator "" _u ## unit(long double m) \
    { return ::operator "" _ ## unit(0.000001 * m); } \
    constexpr T operator "" _m ## unit(long double m) \
    { return ::operator "" _ ## unit(0.001 * m); } \
    constexpr T operator "" _k ## unit(long double m) \
    { return ::operator "" _ ## unit(1000 * m); } \
    // ...

DEFINE_SI_MULTIPLIERS(Length, m)

int main()
{
    constexpr Length l = 3.0_km;
    static_assert(l == 3000, "error");
}

Answer 2

Can't you use the operator "" _m(const char *) flavor as long as you're willing to parse the floats yourself? That makes it possible to write 1234k_m by calling out to a common SI-aware parser for your floating point values.