Why doesn't std::ratio have a value member?

I'm curious as to the reasoning behind why std::ratio only exposes the num and den values, and not some value member which would represent the rational value of the ratio, possibly of the form

static constexpr double value = num / den;

I imagine it could be because

• They might not have wanted to lock in double (or any one particular) precision
• The operation is trivial to perform in user code
• It might require a nasty (unsafe?) cast

but I'm curious if there are other good reasons for the lack of a single value member in the standard?

ASSIDE:

There's a practical application to this for me, because I'm looking to implement a more strongly typed version of std::ratio for a strongly-typed unit library that I've been working on, for which I think something like the following would make sense (assuming I'm not missing some understanding about std::ratio and const-ness)

template<typename UnitType, std::intmax_t Num, std::intmax_t Denom = 1>
struct unit_ratio : private std::ratio<Num, Denom>
{
    static const UnitType value() { return UnitType((double)num / den);
};

In this case, obviously performing num / den in user code is not the same thing as being returned a UnitType which contains the value num / den, but I'm somewhat wary that it's a bad idea since std::ratio doesn't do something similar.