In Programming iOS 4 by Matt Newburg he states:
“To provide a UIViewController with a view manually, implement its
loadView method… You must NOT call super”.
In the iOS 5 Developer's Cookbook by Erica Sadun she states:
“The loadView method allows you to set up the screen and layout any
subviews… Make sure to call [super loadView] whenever you inherit
from a specialized subclass such as UITableViewController or
UITabBarController.”
This, to me at least, is confusing.
Apple is the source of truth and they say NO super call.
If you override this method in order to create your views manually,
you should do so and assign the root view of your hierarchy to the
view property. (The views you create should be unique instances and
should not be shared with any other view controller object.) Your
custom implementation of this method should not call super.
https://developer.apple.com/documentation/uikit/uiviewcontroller/1621454-loadview
[edit]
Another important note scattered around in the UIViewController class reference:
The default loadView method attempts to load the view from the nib
file associated with the view controller (if any).
This is a very old question, but I find that it needs a better answer than the one it got.
Should [super loadView] be called from loadView or not?
It depends. The two sources you cite are talking about different situations, and they're both correct in the context they're describing.
The quote from Neuberg is talking about view controllers that inherit directly from UIViewController. That class has its own implementation of loadView that provides default behavior; specifically, it automatically loads the view hierarchy from a .xib (or storyboard) file that's associated with the view controller. If you call UIViewController's version of that method, the view hierarchy created in that method will either replace your own implementation's view hierarchy, or vice versa. Nine years after this question was posed, the documentation for UIViewController's -loadView method still warns against that:
You can override this method in order to create your views manually. If you choose to do so, assign the root view of your view hierarchy to the view property. The views you create should be unique instances and should not be shared with any other view controller object. Your custom implementation of this method should not call super. [emphasis added]
The quote from Sadun is talking about a different situation, i.e. one in which your view controller is not a direct subclass of UIViewController, but is instead derived from UITableViewController, UITabBarController, etc. Those classes override -loadView themselves and need their versions called. At least in the case of UITableViewController, this is called out in the Overview section:
You may override loadView or any other superclass method, but if you do, be sure to invoke the superclass implementation of the method, usually as the first method call.
So, if you're subclassing UIViewController and providing your own -loadView implementation to generate the controller's views rather than using a .xib or storyboard to provide the views, don't call the superclass's -loadView method. On the other hand, if you're subclassing a class such as UITableView and doing the same thing, check the docs to see whether you need to call that class's -loadView method from your own override.
Related
In a project I'm working on, I have a view controller which overrides the loadView() method for setting up the view hierarchy programatically. I was curious if there are requirements regarding what goes in this method versus in viewDidLoad(), or if the latter is redundant when loadView() is already overridden. Is there anything that cannot be done e.g. at the very end of loadView() that should be deferred until viewDidLoad() is called, or is it simply a matter of preference?
EDIT: The initial phrasing of this question was somewhat opinion based. To clarify, I'm interested in whether there is any reference material indicating that there are technical limitations regarding methods that are not able to be used in loadView() that can be used in viewDidLoad().
The documentation for loadView pretty much answers your question:
If you use Interface Builder to create your views and initialize the view controller, you must not override this method.
You can override this method in order to create your views manually. If you choose to do so, assign the root view of your view hierarchy to the view property. The views you create should be unique instances and should not be shared with any other view controller object. Your custom implementation of this method should not call super.
If you want to perform any additional initialization of your views, do so in the viewDidLoad() method.
In other words, only ever override loadView if you are writing your view controller strictly in code - no nib or storyboard. And then only if you have a reason for the base view of the controller to be something other than a plain old UIView that you get by default. And remember that you must not call super.loadView() in your implementation.
I would use viewDidLoad to create and setup any subviews you wish to add to the base view of the view controller (whether loadView was used or not). Also use viewDidLoad for any other setup that can't be done until the view has been loaded.
Technically, if you override loadView to setup a custom base view, anything you might have put in viewDidLoad can be put in your loadView. But I would still separate the responsibility. Only use loadView (if needed) to create and setup the base view of the controller. Use viewDidLoad to create other subviews as needed and any other controller setup.
Also keep in mind that some view controller initial setup belongs in the initializer. But such code can't make any reference to any views.
Sometimes I want to subclass UIViewController for some app wide customizations. Eg. something that all view controllers should perform during viewDidLoad or viewWillAppear or so.
Naturally I subclass UIViewController and go from there and all view controllers inherit from that. But some of the controllers run tables. And there is UITableViewController designed for that purpose.
So I subclass UITableViewController too and just do the same things there. Which does not seem to be the smartest thing in OOP terms. And there is no multiple inheritance etc.
And as UITableViewController inherits from UIViewController ...
Now I am asking myself why I don't just create my own table view controller that inherits from my very own view controller subclass and adds all the table stuff. But what is "all the table stuff"?
There is the skeleton code that xcode adds to every new table view controller. Quite handy but that can be easily moved to code snippets.
There is the declaration of the protocols UITableViewDelegate and UITableViewDataSource. Easy going. The implementation of those methods has to follow in each subclass of UITableViewController anyway.
There are probably reasonable default implementations of all those mandatory methods in the protocol. Such as returning 0 for numberOfSectionsInTableView: or nil for titleForHeaderInSection or 44.0f for heightForRowAtIndexPath: (Bad example though. Could be smarter not implementing that at all)
So despite the obvious stuff, are there any miracles that UITableViewController takes care of?
I believe all of the behavior UITableViewController adds is well defined in the class documentation: https://developer.apple.com/documentation/uikit/uitableviewcontroller?language=objc
The UITableViewController class creates a controller object that manages a table view. It implements the following behavior:
• If a nib file is specified via the initWithNibName:bundle: method (which is declared by the superclass UIViewController), UITableViewController loads the table view archived in the nib file. Otherwise, it creates an unconfigured UITableView object with the correct dimensions and autoresize mask. You can access this view through the tableView property.
• If a nib file containing the table view is loaded, the data source and delegate become those objects defined in the nib file (if any). If no nib file is specified or if the nib file defines no data source or delegate, UITableViewController sets the data source and the delegate of the table view to self.
• When the table view is about to appear the first time it’s loaded, the table-view controller reloads the table view’s data. It also clears its selection (with or without animation, depending on the request) every time the table view is displayed. The UITableViewController class implements this in the superclass method viewWillAppear:. You can disable this behavior by changing the value in the clearsSelectionOnViewWillAppear property.
• When the table view has appeared, the controller flashes the table view’s scroll indicators. The UITableViewController class implements this in the superclass method viewDidAppear:.
• It implements the superclass method setEditing:animated: so that if a user taps an Edit|Done button in the navigation bar, the controller toggles the edit mode of the table.
All of these are behaviors which should be easy to re-implement if they apply to your specific controller and table view.
In general I have found it preferable to implement these behaviors myself in order to allow for alternate inheritance hierarchies as you noted and because I usually consider setting both the delagate and datasource of a table view to be a view controller a design smell. Those are independent concerns which usually can and should be handled by some other class (e.g. a dedicated data source for a particular model class) rather than bloating a view controller.
So despite the obvious stuff, are there any miracles that UITableViewController takes care of?
Not that I am aware of. As far as I know UITableViewController is mostly a convenience class that can be replaced with your own subclass that adds a few lines of code.
Apple's class documentation pretty much says all that UITableViewController does (and I'm not repeating that here because it may well change in the future). Sometimes, additional information can be gleaned from a class' header file, but in the case of UITableViewController.h the source code comments just repeat what is already in the class docs.
In the end you must decide yourself what you want to replicate in your own subclass. Maybe your project does not need .nib handling? Or none of your table views is editable? So just don't code that...
It seems that the UITableViewController takes care of a lot of management issues that really you could just do for yourself, if you wanted to. Check out the documentation -- it will automatically create a UITableView for you, reload it, etc.
I've gotten pretty comfortable using the viewDidLoad method to execute things I want done at the beginning of a view, but reading one of Apple's tutorials they set the data controller for the class in the awakeFromNib method and did nothing in the awakeFromNib. I swapped it and it seemingly worked identically in my app, but I'm not sure if it was better to have it in awakeFromNib or viewDidLoad.
When should I use either one?
awakeFromNib is called when the associated nib file with a class is loaded . Any class that can own a nib can use it. viewDidLoad is used only by view controllers. It is usually called when loading from nib as well but it can also be called by a view created in memory (very rare circumstance.). If you are using controllers, then I would suggest you to use viewDidLoad
For more Refer this Answer
viewDidLoad is associated with the view controller. If you need to initialize another control unarchived from the nib (e.g. UITableViewCell prototype) you cannot overload viewDidLoad, you need to overload awakeFromNib.
My iOS 5 project uses storyboards. I have a view whose view controller inherits from a class that in turn inherits from UIViewController. This class belongs to a framework that was written for non-storyboard projects, i.e. it initialises the view by overriding loadView. When my view controller runs, it receives self.view as set up by the super class, thus rendering ineffective the Interface Builder settings for the current view.
While I do have the source code of the framework I'm using, and in theory could edit it to remove the loadView override and relocate the code to viewDidLoad for example, I'm wondering if there is a better way to do this just from my own code. Basically I would need to have my [super loadView] un-overriden, or bypass it by calling my superclass's superclass's (i.e. UIViewController) loadView. Is there any way to achieve this?
I thought that I should never call [super loadView] but something is confusing me.
In description of loadView (UIViewController Class Reference) it is said that "Your custom implementation of this method should not call super.",
but in ZoomingPDFViewer example that they gave, loadView implementation (ZoomingPDFViewerViewController) is calling [super loadView].
I have tried to call it from my loadView method and it works ok, but I just don't understand then what does it mean to not call super.
You definitely should not be calling [super loadView]. I'd say you found a bug in the ZoomingPDFViewer example.
You override loadView when you want to programatically create the view hierarchy for your view controller (not using a xib).
As you pointed out, the docs clearly state that you should not call super.
Your custom implementation of this method should not call super.
I assume this is to avoid loading both from a xib and programatically creating a view as this method is used by the base to load a view from a xib:
If the view controller has an associated nib file, this method loads
the view from the nib file.
Note also that even if during allocation of your UIViewController object you pass nil for the nibNameOrNil parameter that the UIViewController implementation of loadView will try to load any xib with the associated class name in it.
A view controller has an associated nib file if the nibName property
returns a non-nil value, which occurs if the view controller was
instantiated from a storyboard, if you explicitly assigned it a nib
file using the initWithNibName:bundle: method, or if iOS finds a nib
file in the app bundle with a name based on the view controller’s
class name. If the view controller does not have an associated nib
file, this method creates a plain UIView object instead.
The real intent of this method is to give you full control of building the view hierarchy without relying on the built in xib loading mechanism.:
You can override this method in order to create your views manually.
Personally, I override loadView if: 1.) The xib I would make for it is really trivial or 2.) The layout of the control is very dynamic, so creating a xib with a static layout has little benefit.
NSViewController tries to initialize a view from a nib in -loadView. Since your nib name is not set for your controller, it will just give you a self.view = nil; I would assume UIViewController works the same way.
So it should be safe, but you it's completely unnecessary.
If you dont have a view created in your IB, then you should call [super loadView] in your code to give a view to your program.
In case of your custom views, you are suppose to create a view with the interfaz builder, so you dont need to call it.
If you create your ViewController programmatically, you could call super.loadView() instead of self.view = UIView(frame: UIScreen.main.bounds) at the beginning of override func loadView().
However, do NOT call self.view before super.loadView(), since the former will trigger the latter if view did not been loaded.
I don't think it's a good idea to explain Apple's documentation like a robot.