When do I initialize a view controller with initWithNibName?

Question

I'm coding in Objective-C for the iPhone. I was wondering when I use init: and when I use initWithNibName:bundle: when creating a view controller.

I can't really find an answer to this question.

Thanks.

Answer 1

Use initWithNibName if you are... initializing with a nib file! That is, a file that you made using Interface Builder.

If you aren't using IB to layout your views, you can just use init.

Answer 2

If you plan on creating your view with Interface Builder then you would use initWithNibName:bundle:. The init function is used if you're going to setup your view hierarchy in code.

A lot of people only use Interface Builder to construct apps, which gets the job done, but I would recommend learning how to do it all in code. IB automates a lot of things and the end result is not always exactly what you want, but if you set everything up in code you have much more control. I've used interface builder very little, and I'm glad I've taken the time to learn how to do lay it all out in code. If you're really in a time crunch IB might be the quicker way, but in the long run manually programming views is worth it.

Answer 3

-initWithNibName:bundle: is the designated initializer for UIViewController. Something should eventually call it. That said, and despite Apple's examples (which favor brevity over maintainability in many cases), it should never be called from outside the view controller itself.

You will often see code like this:

MYViewController *vc = [[MYViewController alloc] initWithNibName:@"Myview" bundle:nil];

I say this is incorrect. It puts implementation details (the name of the NIB and the fact that a NIB is even used) into the caller. That breaks encapsulation. The correct way to do this is:

MYViewController *vc = [[MYViewController alloc] init];

Then, in MYViewController:

- (id)init
{
   self = [super initWithNibName:@"Myview" bundle:nil];
   if (self != nil)
   {
       // Further initialization if needed
   }
   return self;
}

- (id)initWithNibName:(NSString *)nibName bundle:(NSBundle *)bundle
{
    NSAssert(NO, @"Initialize with -init");
    return nil;
}

This moves the key implementation details back into the object, and prevents callers from accidentally breaking encapsulation. Now if you change the name of the NIB, or move to programmatic construction, you fix it in one place (in the view controller) rather than in every place the view controller is used.