This is a simplified version what I would like to do.
constexpr float f(float a, float b){
constexpr float temp = a+b;
return temp*temp*temp;
}
In my version, a+b is something much more complicated, so I don't want to cut and paste it three times. Using 3*(a+b) is also not a working solution for the real function. I'm trying to keep the question related to syntax, and not algebra. I can get it to work by moving a+b to it's own constexpr function, but I'd prefer to not pollute the namespace with otherwise useless functions.