I think the original idea behind compile-time evaluation is that it simplifies metaprogramming (compared to e.g. templates or macros), because there's only one

I think the original idea behind compile-time evaluation is that it simplifies metaprogramming (compared to e.g. templates or macros), because there's only one set of evaluation rules - so the language works the same way, it's just that some parts of the program are run at compile time.