Re: Intercepting retain/release of object

FROM : Stuart Malin
DATE : Wed Mar 19 19:41:30 2008

Thanks all!


On Mar 19, 2008, at 8:04 AM, Jonathan del Strother wrote:
> -(id)retain needs to return itself, not void.



On Mar 19, 2008, at 8:09 AM, Pierre Molinaro wrote:
> The retain method should return self object :


So it should. Changing that fixed the problem.
Hmmm... the compiler never warned me that I had a mismatched return 
type (isn't return type a part of a method's signature?).

I know the release could have freed the object, and so my call to 
retainCount could gen an exception... tolerable in my context of 
debugging.


On Mar 19, 2008, at 8:12 AM, Rob Napier wrote:
> When you say "really odd behavior" can you confirm that you mean
> "actually different from normal behavior" versus "not the behavior
> your were expecting?"


The behavior was different than normal -- my code made the retain 
count value wrong. It is now correct.

>  It does not surprise me that there may be a
> retain involved during the removal of an object from an array. I would
> expect it in order to manage KVO notifications which may fire after
> the object is removed but before it is deallocated.


Yes - that is what I was seeing: a retain upon removing -- which 
confused me.