Question

This is something I should have cleared up long ago, but I just need to know the best practice for deallocating in the following scenario.

In my header file I declare a reference to an IBOutlet as follows:

@interface Test : UIViewController {
    UIButton *_loginBtn;
}

@property (nonatomic, retain) IBOutlet UIButton *loginBtn;

And in the implementation file I associate the instance variable to the property and deallocate as follows:

@implementation Test

@synthesize loginBtn = _loginBtn;

...

- (void) dealloc {
    [_loginBtn release];
    self.loginBtn = nil;

    [super dealloc];
}

- (void) viewDidUnLoad {
    [_loginBtn release];
    self.loginBtn = nil;

    [super viewDidUnLoad];
}

Am I correct in the deallocating the instance variable and setting the property to nil and doing this in both the viewDidUnLoad and dealloc methods?

Answer 1

There is no need for self.loginBtn = nil; in dealloc, the previous line released it. It is best not to use the property to release it in dealloc. The reason for releasing vs setting the property to nil is that the setter is a method call and the class is in the midst of tearing down and things may be unstable.

In viewDidUnLoad release any properties that are IBOutlets with self.theOutlet = nil;, in this case the _loginBtn release]; is not needed and redundant. Also release any other objects that you can easily re-create.

If properties are used they should be used for all accesses in the class with two exceptions: init and dealloc. In both of these cases the class is partially complete. In these two cases it is best to use the ivar directly in init (if necessary) and release in dealloc.