ViewWillDisappear versus dealloc

后端 未结 2 649
伪装坚强ぢ
伪装坚强ぢ 2021-02-10 20:41

I put an NSlog in my dealloc method of my view controller. It does not get consistently called. I do notice that ViewWillDisappear does get called always. Would it be ok to move

相关标签:
2条回答
  • 2021-02-10 20:59

    Well that depends.

    If you need to reduce the memory footprint of your app, unloading stuff while the view is not visible via viewWillDisappear is a good way to go. However you'll need to re-initalize everything once the view will be shown again, depending on its content this may produce considerable overhead - maybe even without the need.

    For a small app (in terms of memory usage), using dealloc to unload your stuff is fine. It will be called only if the objects retain count drops to zero and is the last method that will be run before the object is destroyed. When using autorelease, that may not be the case right away and of course the object could be retained by some object other than the parentviewcontroller, preventing it from being destroyed and thus dealloc from being called.

    You might want to check out the Memory Management Programming Guide, it'll explain things more detailed.

    0 讨论(0)
  • Don't you mean viewDidUnload instead of viewWillDisappear ?

    viewWillDisappear is called when the view controller is disappearing. This usually happens when the view controller is being popped out, or other view controller is pushed to the stack. Purpose of viewWillDisappear is to stop active actions - for example stop animations, hide some elements or similar.

    viewDidUnload is probably what you meant as this one is called when view controller's view is unloaded. This should never happen for a view controller that is currently visible, only for controllers that are somewhere in the navigation stack (part of UITabBarController or UINavigationController), but not currently visible. Purpose of viewDidUnload is to release any UI elements that are part of the view and that the view controller retained as well.

    To understand it's important to realize the reasons why a view for such controller would want to be unloaded. Reason is memory consumption. Views consume considerable amount of memory, even when they are not currently visible. Views are usually very easy to reconstruct - simply by calling the code that constructed them in the first place. That's especially very easy if you are using Interface Builder. So these views are best candidates to be freed to gain more memory.

    When system doesn't have enough memory it starts calling didReceiveMemoryWarning method of view controllers which are not currently visible. If you have created your controllers from template in Xcode, this method simply calls super's (UIViewControllers's) implementation [super didReceiveMemoryWarning]. That default implementation will release self.view, which in turn should deallocate it together with all its subviews.

    Now let's say that your view-controller needs access to some of the subviews to manipulate it in some way. For example you can have a UILabel element there and you want to change its content accordingly. To access this element you create a member variable (IBOutlet) and connect it to the element. Now your view-controller owns that label, so its retain-count is increased. When controller's view is released, so is the label, but because your view-controller still retains the label, it will not be deallocated. Therefore you should release the label in viewDidUnload method.

    I've seen applications that were creating views programmatically (in loadView method), but loading was done in such dirty way that it was not possible to reconstruct the view after it was once deallocated. Therefore each time system was out of memory, it had called didReceiveMemoryWarning which in turn deallocated the view, but after navigating back to that view-controller application had crashed. A fast "bugfix" was to remove calling [super didReceiveMemoryWarning] in view-controllers. Well, system didn't get the memory and some strange effects occurred, but at least the application didn't crash immediately.

    Now the third one - dealloc. This is called when object is not owned by anyone and its memory is going to be freed. Here you need to release all objects that you have retained. For view-controllers those are usually references to model classes.

    I want to describe one more possible scenario. Let's say you have a view-controller displaying a chat with another person. Let's say it's very fancy chat, with emoticons and buddy-icons being animated. Let's say that each chat-entry is displayed as a cell of UITableView.

    When your buddy sends you a message, you want to append a new cell into table-view by reloading it. Therefore your view-controller has an outlet to the table-view.

    In viewWillDisappear you should stop the animations of emoticons and icons. In viewDidUnload you should release the table-view. In dealloc you want to release chat's history (probably NSArray of all messages sent and received during this conversation).

    Now if you navigate away from your chat, viewWillDisappear gets called and you stop the animations.

    When system is low on memory, and your view-controller is not visible, didReceiveMemoryWarning gets called and the view is released. Your viewDidUnload gets called and you release UITableView so that it can be really deallocated.

    When you navigate back to the chat, loadView gets called again and your view is constructed again, viewDidLoad is called after that. Your model (representation of chat conversation) is still there, so the table-view's datasource has all the data as before, so table-view will display exactly the same thing as before the view was deallocated. After all viewWill/DidAppear is called where you start the animations.

    When you finish chatting with your friend, you release the view controller, and it gets deallocated - dealloc is called, you release an array containing the chat messages, and cleanup everything else.

    I hope it make things a little clearer.

    0 讨论(0)
提交回复
热议问题