boost python 中的 enable_pickling 和 pickle_suite 有什么区别？

Question

I have a pretty big class that has ~40 attributes, but only 15 that are passed as arguments to the constructor:

class MyClass{
    /*40 attributes here*/
    MyClass(/*15 args here, the rest are initialized at default values*/);
};

I'm trying to implement pickle support for this class, and so far the boost.python documentation seems to indicate that I can either implement an instance of the pickle_suite class, or I can just do .enable_pickling() and let python handle the rest. Is there any advantage in taking the time to implement all the __getinitargs__ and stuff compared to just doing it with one line? As far as I understand, pickling is essentially converting the objects to a bunch of 1s and 0s and passing it around.

thanks

Answer 1

Nevermind, looks like the normal python pickler can't copy the elements inside a container like a std::vector or a std::pair