When a view controller is first instantiated, it usually creates or
loads objects it needs through its lifetime. It should not create
views or objects associated with displaying content. It should focus
on data objects and objects needed to implement other critical
behaviors.
The above is from the iOS reference :
http://developer.apple.com/library/ios/#featuredarticles/ViewControllerPGforiPhoneOS/ViewLoadingandUnloading/ViewLoadingandUnloading.html#//apple_ref/doc/uid/TP40007457-CH10
The documentation goes on to describe a view load sequence with Storyboard.
My question are :
1
Since a view controller would be associated with a nib file, which contains view objects; And its "viewDidLoad" method seems to be designed for configuring view objects at load time. So how should the documentation's suggestion :
"should not create views or objects associated with displaying content"
be interpreted ?
2
Does question 1 related to whether we use Storyboard or not ?
Not sure I get your question right, but here's my explanation:
initialization and view creation are two separate steps. Let's say I have a view controller with table as IBOutlet which should display a list of recipes stored in core data. In my initialization method I'd fetch the data from CoreData and store it in an array or fetched results controller. I don't need table for that, hence I do nothing with self.view property (which calls the viewDidLoad if there is no view yet). In viewDidLoad I call [tableView reloadData] to redraw the cells so that they display the data from my controller created in controller's initializer.
I don't think it's related, but storyboard should be mere scaffold for your view controllers replacing separate nibs with single file.
The statement you quoted has a lot to do with mobile device limitation and design efficiency. It does not relate to Storyboard in particular.
By "instantiating", the documentation meant the -(id)init; call. When this happens, the controller "prepares critical data, but not creates views". This means the controller evaluates a xib file, and constructs an internal hierarchical representation of the views upon init. This step involves only RAM and CPU.
View controller only creates views when it's pushed into the navigation controller, or view transition animation (that's when viewDidLoad kicks in). This is because views are expensive. It involves GPU and Video RAM. Video RAM is much more limited than RAM, it's not efficient to just create views (back buffer in VRAM) when it's not necessary to display.
If you look at your project, you should discover some view controllers being initialized but are not immediately required to show. Without such design, VRAM will drain quickly for no reason.
Related
Suppose I have 2 views. In the first view, I allocate memory to displaying many UI components such as an UILabel, UIImages, etc.
Suppose the user navigates to the next view (via UINavigationController)
Is it OK to deallocate memory assigned to displaying UI components in the first view and then initialize them again once the user goes back to the first view (in viewFirstLoad or the appropriate function)?
It seems to me if you don't do this, then memory will keep on increasing the longer the user uses your app in that particular session.
Is this not allowed? frowned upon? impossible?
It is perfectly normal and in fact, that functionality is built in standard UIViewController - when controller is not displayed its view may be released from memory and you can release all its subviews (e.g. retained through IBOutlet references) in controller's -viewDidUnload method.
When controller needs to display again it reloads its view again.
It depends. Generally, the rule of thumb is that you should free objects that you don't need. If your view is just a view, then yes, I'd release it and all of its subviews. If your view has data that was obtained through a lengthy retrieval process (e.g. a web service call), I'd probably hold onto the data somewhere so that I don't have to go back out and retrieve it when the user goes back to the first view.
To clarify a little: Apple recommends you display data specific to a view in it's -viewDidLoad method, such as setting text on labels. Then, in -viewDidUnload you should release (or nil outlets of) the view objects you setup in -viewDidLoad. It's critical you implement -viewDidLoad, as the base UIViewController code checks that it's subclass actually implements -viewDidLoad before it assumes it can unload the view (and therefore call -viewDidUnload). Failing to implement -viewDidLoad results in the controller thinking it can't recreate your view at a later time, and so it doesn't unload the view from memory. A developer I know experienced this same problem, took forever to track down.
Any advice on where to store the main list data for an iPhone application like the following?
NavigationController based
Level 1 (main screen) is a List of Items. Hence it uses a table view (table of items)
Level 2_EDIT: Is a view you can get to from the main screen by clicking EDIT. Here you can add text to be added to the main view list
Level 2_DETAIL: Is a view you can get to from the main screen by clicking on a cell.
Now assuming the implementation is (rough overview):
* MainView - appDelegate (holds UIWindow & UINavigationController)
* RootViewController - table view of the main items list (? variables here ?)
* EditViewController - input text to add to main list
* DetailViewController - shows detail of record
Question - Where to hold the NSArray that keeps the main list of items? Should it be in the RootViewController where the Table View exists that displays it? Or should it be higher up in the ApplicationDelegate? I note that when you go from RootViewController to EditViewController, then in this edit view you would have to ADD items to the array, so would it be easier for the code in the EditViewController to access the main array from the AppDelegate (as opposed to the RootViewController)?
(Note - still haven't made an app that has a specific model object, re MVC, yet, so not sure if this should come into the picture.)
Your data becomes the table view's data source, and although it doesn't have to, the view controller that contains the table view often does the data source's role. The view controller can also be a delegate for the EditViewController, so the EditViewController sends a message to it so it can update the array.
Apple's CoreDataBooks sample project show similar architecture. You may want to take a look.
Having the array in the application delegate is often not a great idea. Although it can give you a little bit of convenience, now your classes totally depend on your application delegate unnecessarily.
Your table view shows your data. This corresponds to View in the MVC design pattern. I assume RootViewController is the view controller of the table view, which acts as a Controller in the pattern. Your data, the location of which is not decided yet, corresponds to Model. The role of RootViewController becomes connecting the Model and View.
The ideal, or the reason of the MVC pattern is to isolate the model and the view, so the model can work with other views with appropriate controllers, and the view can also work with other models with appropriate controllers. For example, your RootViewController will provide the table view with data. It will specify the data in the language of table view, e.g. the number of sections and rows, the contents of cells, etc. If you want to present the data in a different way, for example a graph, your controller will access the same data (model) and provide the graph view with a different representation of the same data. The model need not change, neither do the views. You write only the controller, for each combination of a model and a view.
Ideally, therefore, you will have a different class for the model. In this class, you will store the array, and provide a general interface for controllers can interact with the data.
However, it's often not that necessary, either because it's unlikely that you will use the model class often again, or because the model itself is way too simple so it can be easily implemented anywhere. For example, if your data for the table is a simple array, an NSArray object is often sufficient for the model's role. Therefor, here comes an idea that to combine the controller and the model into a single object.
This is why it makes sense that your table view controller often acts as the data source of the table view.
However, storing the data in the application delegate is a completely different idea. Now the application delegate becomes your model, but which does not make sense, because the application delegate is only used for the specific application. Why would you have a separate model object that totally depends on a single application? Also, if your table view directly interacts with the application delegate, it means that now your view cannot work for other applications either because it now depends on the specific application's application delegate.
Often the reason why people are tempted to have data in the application delegate is, that the application delegate is easily accessible by any objects in the application by using [UIApplication sharedApplication].delegate. The M-V-C relationship is not always very simple. For example, your EditViewController also need to access the same model. To do this, you have to write some code to make the model accessible by both the table view and the edit view. If you have the data in the application delegate, you don't need to do anything, because you can magically access the array by accessing the application delegate.
But that's all. A few minute's saving in your coding time, for the price of ruining your software architecture. I'm not a fundamentalist, and I do sometimes use application delegate to store some data when I'm absolutely sure that it's not worth to provide well-formed interfaces, but it's rare.
So how should you connect your edit view to the data merged into the table view controller? There could be multiple ways. What I suggested before is let the edit view controller has a weak reference to the table view controller (delegate) and send a defined message, for example - (void)editViewController:(EditViewController *editViewController) didFinishEditing:(id) someData. In this way, you can use this edit view controller with some other view controllers as long as they use the same protocol. But others may implement different interfaces for it.
You need to store in the appDelegate,as you can acees the data in anyviewcontroller then.But if you use rootviewController,then you can only send data to next ViewController then,not to the 2nd or 3rd ViewController directly.
Hope it will help you.
Good Luck
I'm currently working on an app that uses a UINavigationController inside UITabBars. The tab bars correspond to both UITableViews as well as a Map View. However, the root view controller of the app is not the parent, or direct parent, of the UITableView custom controllers and map view controller. I also have a p-list that creates NSDictionary objects; it is the datasource that I am using to populate entries in the tables and the map.
So, without a root view controller, how should I create the NSDictionary objects from the data source? It seems to me that the easiest way is to simply recreate the object in each view controller for a view that needs the data. Because I don't have that many views and the p-list isn't that long, memory shouldn't be an issue. However, I do know that this is all very inefficient.
Is there any alternative implementation so that I don't have to recreate the NSDictionary in each view controller?
This tutorial featuring the singleton guide neatly explains everything:
http://www.cocoanetics.com/2009/05/the-death-of-global-variables/
My only worry now is if multiple view controllers each call the singleton, wouldn't they all be generating multiple instances of the NSDictionary, and couldn't that conceivably still take up a lot of memory?
You need a data model object that stores the data for application.
A data model is a customized, standalone object accessible from anywhere in the application. The data model object knows nothing about any views or view controllers. It just stores data and the logical relationships between that data.
When different parts of the app need to write or read data, they right and read to the data model. In your case, view1 would save its data to the data model when it unloads and then view2 would read that data from the data model when it loads (or vice versa.)
In a properly designed app, no two view controllers should have access to the internal data of another controller. (The only reason a view controllers needs to know of the existence of another controller is if it has to trigger the loading of that other controller.)
The quick and dirty way to create a data model is to add attributes to the app delegate and then call the app delegate from the view controllers using:
MyAppDelegateClass *appDelegate=[[UIApplication sharedApplicaton] delegate];
myLocalProperty=appDelegate.someDataModelProperty;
This will work for small project but as your data grows complex, you should create a dedicated class for your data model.
Placing the data in an init method or a viewDidLoad won't work because in a tabbar the users can switch back and forth without unloading the view or reinitializing the view controller.
The best place to retrieve changing data is in the viewWillAppear controller method. That way the data will be updated every time the user switches to that tab.
In practical terms, a data model is usually either placed in some kind of a singleton (either your own, or the app delegate which is a kind of singleton).
You should read up on Model-View-Controller (MVC) architecture. Specifically, you'll probably want to introduce a data model where you would create and populate the dictionary during initialization and then access it from all of your view controllers.
I want to access the value of a view controller data member variables in another view controller object.
Or is it possible to access its controls like UILabel text property?
A lot of times when I find I have to do things like that I find I can redesign the solution and the need for it goes away. Jay's law: "If it's too hard you're probably doing it wrong."
It is possible to access a UILabel of another view controller, but don't. It will lead you to very hard-to-understand bugs. Any IBOutlet can become nil at surprising times when memory is low. You shouldn't mess with another object's UI elements directly.
Your initial idea of accessing the data (model) objects, is the right one, though generally you will be better off to just initialize both view controllers with the same model object. For instance, say you have a status message that you want to show up in two different UILabels in two different view controllers. Rather than have one view controller ask the other view controller for the data, it's better to have a model class like "Status" that both views have a pointer to. Whenever it changes, they change their UILabel.
Even better is to post a notification (StatusDidChangeNotification) and just let everyone who cares observe it and update their UI appropriately.
You want to keep UI elements very loosely coupled in Cocoa. Otherwise you wind up with hard-to-fix bugs when you make what seems like a minor UI change.
You are going to have to define the property in the view controllers interface, then as long as you have a reference to the view controller in the second view controller you should be able to access it like the text of a UILabel..
viewWillAppear: is only called by the framework when you use the built-in view controller transitions like presentModalViewController:animated: or pushViewController:animated:. In other cases, you have to call viewWill/Did(Dis)Appear: yourself.
I have a view that contains a UITableView and a UILabel which works perfectly as far as I can tell. I really don't want to manage the UIView and UITableView with the same controller as the UITableViewController handles a lot of housekeeping and according to the documentation:
If the view to be managed is a
composite view in which a table view
is one of multiple subviews, you must
use a custom subclass of
UIViewController to manage the table
view (and other views). Do not use a
UITableViewController object because
this controller class sizes the table
view to fill the screen between the
navigation bar and the tab bar (if
either are present).
Why does Apple warn against using it and what will happen if I ignore this warning?
Update: Originally I quoted the following from the Apple Documentation:
You should not use view
controllers to manage views that fill
only a part of their window—that is,
only part of the area defined by the
application content rectangle. If you
want to have an interface composed of
several smaller views, embed them all
in a single root view and manage that
view with your view controller.
While this issue is probably related to why UITableViewController was designed to be fullscreen, it isn't exactly the same issue.
The major practical reason to use only one view controller per screen is because that is the only way to manage navigation.
For example, suppose you have screen that has two separate view controllers and you load it with the navigation controller. Which of the two view controllers do you push and how do you load and reference the second one? (Not to mention the overhead of coordinating the two separate controllers simultaneously.)
I don't think using a single custom controller is a big of a hassle as you think.
Remember, there is no need for the TableviewDataSource and the TableViewDelegate to be in the actual controller. The Apple templates just do that for convenience. You can put the methods implementing both protocol in one class or separate them each into there own class. Then you simply link them up with the table in your custom controller. That way, all the custom controller has to do is manage the frame of tableview itself. All the configuration and data management will be in separate and self-contained objects. The custom control can easily message them if you need data from the other UI elements.
This kind of flexibility, customization and encapsulation is why the delegate design pattern is used in the first place. You can customize the heck out of anything without having to create one monster class that does everything. Instead, you just pop in a delegate module and go.
Edit01: Response to comment
If I understand your layout correctly, your problem is that the UITableViewController is hardwired to set the table to fill the available view. Most of the time the tableview is the top view itself and that works. The main function of the UITableViewController is to position the table so if you're using a non-standard layout, you don't need it. You can just use a generic view controller and let the nib set the table's frame (or do it programmatically). Like I said, its easy to think that the delegate and datasource methods have to be in the controller but they don't. You should just get rid of the tableViewController all together because it serves no purpose in your particular design.
To me, the important detail in Apple's documentation is that they advise you not to use "view controllers [i.e., instances of UIViewController or its subclasses] to manage views that fill only a part of their window". There is nothing wrong with using several (custom) controllers for non-fullscreen views, they just should not be UIViewController objects.
UIViewController expects that its view takes up the entire screen and if it doesn't, you might get strange results. The view controller resizes the view to fit the window (minus navigation bars and toolbars) when it appears, it manages device orientation (which is hard to apply correctly if its view does not take up the entire screen) etc. So given how UIViewController works, I think there is merit to Apple's advice.
However, that doesn't mean that you can't write your own controller classes to manage your subviews. Besides the things I mentioned above, interacting with tab bar and navigation controllers, and receiving memory warnings, there isn't really much that UIViewController does. You could just write your custom controller class (subclassed from NSObject), instantiate it in your "normal" fullscreen view controller and let it handle the interaction with your view.
The only problem I see is the responder chain. A view controller is part of the responder chain so that touch events that your views don't handle get forwarded to the view controller. As I see it, there is no easy way to place your custom controller in the responder chain. I don't know if this is relevant for you. If you can manage interaction with your view with the target-action mechanism, it wouldn't matter.
I have an application where I did use 2 separate UIViewController subclasses below another view controller to manage a table view and a toolbar. It 'kind of' works, but I got myself into a massive pickle as a result and now realize that I should not be using UIViewController subclasses for the sub controllers because they contain behavior that I don't need and that gets in the way.
The sort of things that went wrong, tended to be:
Strange resizing of the views when coming back from sub navigation and geometry calculations being different between viewWillLoad and viewDidLoad etc.
Difficulty in handling low memory warnings when I freed the subview controllers when I shouldn't have done.
Its the expectation that UIViewController subclasses won't be used like this, and the way they handle events, using the navigation controller etc that made trying to use more than one UIViewController subclass for the same page tricky because you end up spending more time circumventing their behaviour in this context.
In my opinion, The Apple Way is to provide you the "one" solution. This served the end-users very well. No choice, no headache.
We are programmers and we want to and need to customize. However, in some cases, Apple still doesn't want us to do too many changes. For example, the height of tab bar, tool bar and nav bar, some default sizes of the UI components(table view), some default behaviors, etc.. And when designing a framework and a suite of APIs, they need to nail down some decisions. Even if it's a very good and flexible design, there is always one programmer in the world wants to do something different and find it difficult to achieve against the design.
In short, you want a table view and a label on the same screen, but they don't think so. :)