[Pyrex] Please support C++

[Lenard Lindstrom]

Thanks for the information. I may leave things as they were. I've just been studying 
Coercion nodes and they look like a better solution than what I am doing now. Well, 
part of my solution will be salvaged.

Lenard Lindstrom
<len-l at telus.net>

On 18 Oct 2005 at 14:38, Greg Ewing wrote:

> Lenard Lindstrom wrote:
> 
> > Pyrex has code to 
> > ensure all such pointers are (PyObject *). I took out that restriction to allow 
> > temporary pointers to extension types to be of the type of the extension type. 
>  >
>  > Does removing the restriction re-introduce some kind of bug?
> 
> It looks like the code for managing temps of different
> types is intact, so there shouldn't be any problem.
> 
> As far as I remember, the main reason for making all the
> temps PyObject * was that they are very frequently passed
> to Python/C API functions, and doing things that way
> minimised the total amount of casting going on in the
> generated code, making it easier for me to see whether
> it was correct.
> 
> Another reason is that it maximises the amount of
> re-use of temps that is possible, although that's
> probably not a big concern.
>