Summary: How should the UIViewController
know the size of its UIView instance when initialising that view?
The dedicated initialisation method for an UIView is the initWithFrame:(CGRect)frame
method. This sets the frame for the newly created UIView. This method could be called from the UIViewController's
loadView method, if that view controller's view is requested. The documentation of UIViewController
states with respect to subclassing and view size:
When creating the views for your view hierarchy, you should always set the autoresizing properties of your views. When a view controller is displayed on screen, its root view is typically resized to fit the available space, which can vary depending on the window’s current orientation and the presence of other interface elements such as the status bar.
So, the UIViewController
instance should set those properties. So far so good, the UIViewController
so far does not have to know how big its view is or will be.
When the view of a UIViewController
is requested and the view property is nil, the loadView method of the view controller is called. Now there is a problem, because the UIView
needs to be initialised, but the view controller still does not know what size the view should be. How big should you initialize that view? Where in code do you determine the view size?
You could initialize the view with a zero rect (CGRectZero
):
- (void)loadView {
self.view = [[[UIView alloc] initWithFrame:CGRectZero] autorelease];
}
And let the caller set the view frame like so:
UIViewController *viewController = [[MyUIViewController alloc] init];
// next two lines are normally combined, but for clarity they are not now
UIView *view = viewController.view;
view.frame = CGRectMake(0, 0, 200, 200);
This requests the view from the view controller (viewController.view), and thus loads its view with the loadView method. This loadView method initializes the view with a CGRectZero
. Then the caller sets its frame (view.frame = ...)
The thing is that the frame property on the view is set twice, possibly causing even more double work if your custom UIView
is doing some advanced layout in the setFrame method (placing and resizing subviews for example). You could prevent this by creating a dedicated initializer method for the UIViewController
which already asks the caller for a CGRect, which you would store in an ivar. At the time the loadView method is called, you use this ivar to create the view.
What is the good way to go here? Either setting the view's frame twice (initializing with CGRectZero
, and setting afterwards), or giving the UIViewController
a new initializer method with a CGRect (and thus giving it a frame property)? Or am I missing something and are there other possibilities?