[PyKDE] Re: Internal Pointers again... Was: Model indexes and internal pointers

[Mike Rovner]

Phil Thompson wrote:
>>>> unique as far as python is concerned.  So when I put this into
>>>> internalPointer to retrieve later it immediately goes out of scope in
>>>> python and is invalid when I try to reference it later.

>> PyQt can return new wrapper every time but all of them pointing to the
>> same C++ object, can it?
> Not at the moment, but SIP could be changed to make this an option for
> specific classes. However, I don't see how it helps - when the C++ object
> disappears the Python object won't know and will crash when referencing
> it.

Right, that can't prevent referencing disappeared C++ object.
However as I understand OP problem, he got Python wrapper goes out of 
scope while he is keeping C++ ref to it.

My (not mine actually, see boost.python return_internal_reference 
policy) is to keep C++ object and keep C++ ref to this obj in client 
prog. SIP will generate a new wrapper obj referencing same C++ obj, so 
when wrapper goes out of scope it's destroyed but not the C++ object it 
wraps.

/m