Re:PyCXX question on __module__ attribute

[kishor s. <kis...@gm...>]

>Its been a long time since I used pickle. Personally I would avoid pickle
because of the security and robustness issues.

Anyway. It looks like if you want to pickle an extension object you have to
implement the
__reduce__ and/or __reduce_ex__ methods. You would have to see what pickle
does with the tuple returned by __reduce__ to recreate the objects. You can
also register functions to assist pickle.

Given that the reason a C++ extension module is created is to wrap around
complex resources I wonder if
its reasonable to attempt to pickle a C++ object.


Thanks Barry.  If we leave pickling issue for a while, do you know why
__module__ is not when we create the extension using PyCXX though we pass
the module name using below

    example_module()
        : Py::ExtensionModule<example_module>( *"example"* )


Most of the python extension module have __module__ set which PyCXX
extension doesn't.


In [4]: cPickle.load.__module__

Out[4]: 'cPickle'



Any thoughts?

Thanks,
Kishor