Question

I have several objects in my app that can become nil at some point and I have methods that in theory are used to put these objects to nil.

But, if I try to put to nil an object that does not exist, the app will crash.

for example…

[object1 release];
object1 = nil;

//… and after that

[object1 removeFromSuperview]; // this will crash

Then I thought, why not testing to see if the object exists before removing…

if (object1 != nil)
 [object1 removeFromSuperview]; 
// this will crash too, because object1 cannot be tested for nil because it does not exist

How can I check if the object exists before testing if it is nil?
something as

if (object1 exists( {
  if(object1 != nil))
    [object1 removeFromSuperview)
}

is this possible?

I ADDED THIS TO CLARIFY…

what I mean is this: imagine I have object1 declared on the header and initialized on the code. So, it exists and points to a valid reference. At some point, the object may be released, so the reference still points to an object but the object was freed. Even if I put the object to nil after releasing, I cannot do anything with it.

The problem is this: I have some methods that are asynchronous. One of them scans for some objects and remove them if they are found. I must check if the object exists and the reference points to a valid object before releasing it again. This is the point: how do I test if the object exists and its reference points to a valid existent object before releasing it again, to void releasing again an object that is already released and crash the app.

Answer 1

In short, your question of determining if the pointer is still valid is going down the wrong path.

After you release an object, you must immediately set its value to null so that you no longer control it. If two methods are running asynchronously accessing the same object, they must be synchronized at that point so that releasing and setting to null happens at the same time in one thread before the other has a chance to interrupt.