Question

What reasons are there for doing in three lines what could be done in one?

Here is some code from developer.apple.com:

UINavigationController *aNavigationController = [[UINavigationController alloc] initWithRootViewController:rootViewController];

self.navigationController = aNavigationController;

[aNavigationController release];

…and the same thing in one line:

self.navigationController = [[UINavigationController alloc] initWithRootViewController:rootViewController];

It seems clean, simple, and straightforward enough. I’ve had trouble in the past with a property not being retained, causing [object release] to destroy the object when it wasn’t supposed to (so far as I could tell – the retain attribute was set). Using the one-line formula works like a dandy.

Answer 1

The Objective-C memory management rules dictate that by allocing an object instance you are a (shared once other objects retain it) owner of that instance and so you must release the UINavigationController when you want to relinquish ownership to prevent a memory leak. In a non-garbage collected environment (e.g. on the iPhone) this means balancing alloc or copy (or methods that contain “alloc” or “copy”) with release or autorelease. Your second snippet would then be

self.navigationController = [[[UINavigationController alloc] initWithRootViewController:rootViewController] autorelease];

If you can avoid using autorelease on a memory-limited environment (like the iPhone), it’s better to use explicit release. -[NSObject autorelease] adds the receiver to the current NSAutoreleasePool which will subsequently call -release on the objects in the pool ‘some time in the future.’ When you want to be careful about memory usage, ‘some time in the future’ is not a good idea. Thus, on the iPhone your first example is the standard usage.