Re: [Pyobjc-dev] PyObjC 1.1b1 released

[Ronald O. <ous...@ci...>]

On 23-feb-04, at 20:32, Pierce T. Wetter III wrote:

>
> On Feb 21, 2004, at 1:03 PM, Ronald Oussoren wrote:
>
>> I've just pushed release 1.1b1 to the website.
>
>   What could I do to convince you to add the
>
>  [obj self]
>
>   call to the release? Its the only change I have left to make pyobjc 
> work well with EOF 4.5, and while its a no-op for non-EOF objects, its 
> necessary to trigger faults before converting to a pyobj. Its really 
> annoying to have to distribute a custom version to our developers.

In august 2003, Pierce wrote:

>  I found one spot:
>
> Changing
>
> #define   PyObjCObject_GetObject(object) 
> (((PyObjCObject*)(object))->objc_object)
>
> to
>
> #define   PyObjCObject_GetObject(object) 
> ([((PyObjCObject*)(object))->objc_object self])
>
> in pyobjc.h seems to do it.

Are you still using this patch?

If so, could you try if adding 'objc_object = [objc_object self]' just 
below the code below (in Modules/objc/objc-object.m, line 712, in 
function PyObjCObject_New) also works?

         res = find_existing_proxy(objc_object);
         if (res) return res;

If I recall the issues correctly this should convert any "ghost" 
objects to real objects before passing them to python. This reduces the 
changes of confusing the bridge w.r.t. the patch above.

A simular change might be needed in PyObjCObject_NewUnitialized, that 
function is only called for the result of +alloc and probably doesn't 
need changing.

I'll add this to the bridge if this solves your problems with EOF (and 
doesn't introduce new problems).

Ronald
--
X|support bv            http://www.xsupport.nl/
T:  +31 610271479       F:  +31 204416173