Currently there is an explicit T operator()() const to retrieve the underlying value.
I would like to get access to the underlying value as a non-const reference to modify it : explicit T& operator()()
Motivation:
I would like to reuse some numerical utils methods that take doubles e.g.
bool my_method(double& a, double b){
// modifies a
}
I could template my_method on a unit type but I am concerned with two things:
1) it is intrusive - I would like to avoid modifying the existing code if possible
2) (more importantly) it will lead to different template instantiations for each unit type I am using and I must keep binary size to a minimum (embedded context).
So I thought I could do:
template<typename U>
bool my_method(U& a, U b){
return my_method(static_cast<double&>(a), static_cast<double>(b));
}
which the compiler should be able to compile away entirely, giving me zero-compiled-size-and-runtime-cost
Potential issue: that would work only for linear_scale, not for DB scale (but I am only using linear).
I have a simple change for that, would you consider a PR for that feature?
Currently there is an
explicit T operator()() constto retrieve the underlying value. I would like to get access to the underlying value as a non-const reference to modify it :explicit T& operator()()Motivation: I would like to reuse some numerical utils methods that take doubles e.g.
I could template my_method on a unit type but I am concerned with two things: 1) it is intrusive - I would like to avoid modifying the existing code if possible 2) (more importantly) it will lead to different template instantiations for each unit type I am using and I must keep binary size to a minimum (embedded context). So I thought I could do:
which the compiler should be able to compile away entirely, giving me zero-compiled-size-and-runtime-cost
Potential issue: that would work only for linear_scale, not for DB scale (but I am only using linear). I have a simple change for that, would you consider a PR for that feature?