- An integer number can be translated to a floating point number (e.g., by calling `get<double>()`).
- A floating pointnnumber can be translated to an integer number (e.g., by calling `get<int>()`). Note the number is truncated and not rounded, ceiled or floored.
- Any value but JSON objects can be translated into an array. The result is a singleton array that consists of the value before.
- Any other conversion will throw a `std::logic_error` exception.