[C++-sig] shared_ptr trouble

[Alex Mohr]

>   Instead, it creates a
>   new shared_ptr (with a whole new "control block" containing a
>   use_count, weak_count and deleter) whose deleter manages a reference
>   count on the owning Python object.  That's the magic that allows
>   Boost.Python to recover the original Python object when such a
>   shared_ptr is converted back to Python.

I've been meaning to post regarding this and my effort to make custom 
smart pointers work as well as shared_ptr does with c++ <-> python 
polymorphism.

This "object identity" problem is the biggest issue.  You can sort of 
make it work with custom smart pointers by registering from_ and 
to_python conversions that store and lookup a handle to the associated 
python object.  It doesn't completely work, because class_<> registers a 
to_python conversion for you.

I'd love to have a generic mechanism where users could customize the 
object identity mapping.  Perhaps something similar to get_pointer 
(set/get_object)?  Ideally we could use this to remove some of the 
special-case shared_ptr code, and just provide a default implementation 
for shared_ptr.

If this seems like a good idea, I'd be willing to work on this when I'm 
finished with the (*args, **kw) stuff.

Something like this may be useful to help solve Ralf's problem. 
Actually, if one is willing to have an "invasive" handle to the python 
object stored in the c++ object, I think it's definitely solvable.

Alex