Fully functional std::complex specializations and overloads

[ChrisChiasson]

@rok_cesnovar

I dont want to derail this topic but not sure what you mean by “not working”? In order to do dynamic linking on Windows TBB needs to be in the users or systems PATH variable.

rpath, if I understood it correctly, is different from the PATH environment variable. I believe you’re correct about PATH (in that it is one of the ways to solve the problem). I had only mentioned rpath because it is in the instructions for how to compile a stan math program in the readme.md.

@ChrisChiasson

I’ll check when I get in to work tomorrow, but it was telling me that pow(var,int) (I think that’s what it said) was ambiguous, an error that didn’t show up when using my implementation.

It was pow(var,int). The beginning of the error snippet was as follows:

src/Nautical/MotionTables/MotionTables.hpp:55:16: error: call of overloaded 'pow(stan::math::var, int)' is ambiguous
   55 |        ess+=pow(ERSAft-check,2);
      |             ~~~^~~~~~~~~~~~~~~~
compilation terminated due to -Wfatal-errors.
In file included from src/main.cpp:9:
./src/Nautical/MotionTables/MotionTables.hpp:55:13: fatal error: call to 'pow' is ambiguous
       ess+=pow(ERSAft-check,2);

@Bob_Carpenter

CRTP base class

I have been wondering, why are you calling it CRTP? The pattern I see is

template <>
class complex<stan::math::var>
    : public stan::math::complex_base<stan::math::var> {

@Bob_Carpenter

And I wanted to support mixed operations of things like std::complex<double> and std::complex<var> without promotion.

I agree with that and consider it worthwhile. I would only note the approaches are not mutually exclusive. In my implementation, sometimes we selectively avoided delegating back to the standard library, especially with g++ (as you mentioned), by writing declarations with more specific types to give them higher priority in template selection and overload deduction.