Followup: Weakref - details

[ATS]

> 
>  I wonder if it would also work if we use "class __multiple_inheritance
> wxEvtHandler;" when we forward-declare it? It would be great if you could
> test this as it seems to be like the least intrusive workaround.
> 
> A> I don't use eVC but was guessing that the same would be possible there.
> 
>  From eVC help file it does support these options and it also supports
> #pragma pointers_to_members(multiple_inheritance) and
> __multiple_inheritance keyword.

The thing is, with eVC you would be casting to a "pointer to wxObject member
function". That's a single inheritance class. 

When we use a "pointer to wxEvtHandler member function", we get a more 
general pointer (larger), so we can handle both single & multiple 
inheritance. By moving to wxEventFunction we are prepared for both 
cases (and option "/vmm /vmg" sets all member function pointers to 
this larger size).

Using wxObjectEventFunction for eVC, I can't see its possible to avoid
the extra compiler options. (It's like trying to fit an int into a short).



> A> >  I must be missing something but what message exactly will he get?
> A> 
> A> Weak refs untested on eVC and that "/vmm /vmg" could help getting it 
> A> to work. 
> 
>  What I meant to ask was where would this message be coming from. If we add
> it to wx/event.h as a #pragma message it will be triggered by all code
> which will be worse than annoying. Do you see a better alternative?

How about setup.h, as a comment, where the wxUSE_WEAKREF option is defined?
Or if there is a port specific README.TXT (for WinCE). 


Also, are you sure this compiler error still exists? I was searching for
some mention of it elsewhere but with no result. 

It's a compiler bug, and ought to be fixed.


Regards
// ATS


> 


>  Thanks,
> VZ
>