When I simulate a memory warning, viewDidUnload should run on unused objects, right?
How do I go about figuring out WHY my UIView won't go away?
FYI I'm using ARC and every ivar is an IBOutlet and looks like:
#property (nonatomic, weak) IBOutlet UIView *someView;
What class are we looking at here? Only UIViewControllers release their view in case of a mem warning.
If this is a custom class or a custom added view, you should unload it yourself.
Related
When I control-drag outlet from the storyboard, I got the strong property by default. I'm using Xcode 9.4.1. The following code is what it looks like. Is it because of XCode Swift compatibility as in Swift it makes sense to create strong.
#property (strong, nonatomic) IBOutlet UILabel *lbl
The default value should be weak, but if you once change it manually to strong as shown in the photo, the default value changes to strong, so you need to manually put it back to weak.
I am using XCode 5.1.1, targeting iOS 7.0.
When creating outlets from my storyboard using the Assistant editor. I notice I have a few choices to create properties or ivars. The one I have been using is dragging directly to my *.m #implementation and it creates code like:
#implementation AudioViewController
{
__weak IBOutlet UILabel *posLabel;
__weak IBOutlet UILabel *durationLabel;
__weak IBOutlet UIButton *playButton;
}
I have no need to access these outside of this class, so this seems convenient, but I am wondering if there are any "gotchas" to this method vs creating properties, especially in regards to memory management. I read on other stack answers that you must create (weak) properties or I will have to [release] manually. I am wondering if this __weak takes care of that in this context?
Thanks!
Creating properties and instance variables with the same modifier is mostly analogous. When you are using ARC, you do not have to release strong properties or instance variables - they will be released when the object is deallocated. Interface element outlets are usually created as weak, because they are retained by the view hierarchy. You should be careful; if you intend to remove the elements from the view hierarchy at some point, you should change the modifier to strong to ensure they are retained by the view controller. Top-level outlets should also be created as strong to make sure they are retained after nib load.
I'm really struggling to understand memory leaks from a custom tableViewCell in an ARC-based project and was hoping someone might be able to shed some light on if the following might be responsible or if I'm barking up the wrong tree entirely.
So, I'm pushing a tableViewController into a navigationController and in the tableView I have my own custom tableViewCells.
This custom tableViewCell has a delegate property which is assigned from cellForRowAtIndexPath and points to the tableViewController itself, so in other words...
cell.delegate = self;
This custom tableViewCell also has a couple of observers for NotificationCenter.
Now, I've put in code to remove the observers in the tableViewCell dealloc and likewise went to add...
self.delegate = Nil;
...only to discover dealloc doesn't ever seem to be called, even when going tapping Back (i.e. going back up the navigation stack effectively finishing with the pushed tableViewContoller).
So, I guess this boils down to: could either of these aspects of the cell cause a leak? And if so, am I going about stopping the leaks in the right way?
EDIT: I'm defining my delegate property as follows:
#property (nonatomic, retain) MyTableViewController *delegate;
You need to declare your custom delegate as weak
#property (nonatomic, weak) id<MyCustomDelegateProtocol> delegate;
Otherwise you get a strong reference cycle
controller -> table view -> cell -> controller
n.b. weak is specific to ARC, if you're not using ARC you need to use assign, and make sure to nil it out yourself when you're done being its delegate.
Another issue can be this one:
iOS 7.0 and ARC: UITableView never deallocated after rows animation
Not applicable if you're running on iOS 7.0.3+.
Previously I've been creating my views with interface builder.
When creating views with storyboards or nibs I would connect my outlets. I understand that the outlets are creating a strong connection to the properties on the view.
If I am creating views programmatically should my properties be strong?
.h
#interface LoginViewController : UIViewController <UITextFieldDelegate>
#property (nonatomic, strong) UIView *loginView;
#property (nonatomic, strong) UITextField *usernameTextField;
#property (nonatomic, strong) UITextField *passwordTextField;
#property (nonatomic, strong) UIButton *signInButton;
#end
.m
#interface LoginViewController ()
#end
#implementation LoginViewController
- (void)initViewsAndLayout
{
_loginView = [[UIView alloc] init];
_loginView.frame = self.view.bounds;
[self.view addSubview:_loginView];
//...
}
#end
A few things:
Your code isn't using the properties you defined
Don't put the private properties in the .h file
Don't state that your class conforms to the UITextFieldDelegate protocol in the .h file
As for whether the properties should be strong or weak I prefer strong but since you will be adding each of these properties (views) to the view controller's view, it would be fine to make them weak since there will always be a reference to them as long as the view controller is alive.
Your .h should just be:
#interface LoginViewController : UIViewController
#end
Your .m should be (assuming you do want to use the properties):
#interface LoginViewController () <UITextFieldDelegate>
#property (nonatomic, strong) UIView *loginView;
#property (nonatomic, strong) UITextField *usernameTextField;
#property (nonatomic, strong) UITextField *passwordTextField;
#property (nonatomic, strong) UIButton *signInButton;
#end
#implementation LoginViewController
- (void)initViewsAndLayout
{
self.loginView = [[UIView alloc] init];
self.loginView.frame = self.view.bounds;
[self.view addSubview:self.loginView];
//...
}
#end
When you use the strong attribute, you are basically writing retain, as in, you want Objective-C to allocate memory for the property and hold it until it is released.
That is all strong means.
This is only necessary if you are using ARC (though you can still use retain in ARC). If not, use retain.
"If I am creating views programmatically should my properties be strong?"
If they are objects and not primitives, then yes.
I recommend you to make properties with strong attribute if you want to create it manually not to use xib or storyboard.
As your code in initViewsAndLayout, if you always create a subview and add it to a view of a viewcontroller simultaneously in same method scope, there is no problem even if you use weak or assign attribute. addSubview will increase a reference count of the subview. But, I don't think it's such a good habit because all of us could make a mistake.
Therefore, you had better make a property of UIView with strong attribute.
You asked:
If I am creating views programmatically should my properties be strong?
Two part answer:
The view controller should definite maintain a strong reference to its top level view.
When not using NIBs or storyboards at all and doing everything programmatically, the root view is instantiated in loadView and generally stored in a retain/strong property. UIViewController already has a view property that bears the retain memory semantics, thats generally used. (For more information about programmatically created views, see the Creating a View Programmatically section in the Resource Management in View Controllers chapter of the View Controller Programming Guide for iOS)
(If you're not creating the root view in loadView and are instead instantiating the root view with a NIB or storyboard and are only programmatically creating the subviews, don't worry about the above discussion, as the NIB/storyboard takes care of all of that for you.)
For the subviews, when you call addSubview, the view is retained by its parent view. It's unnecessary for the view controller to also maintain strong reference to it as well. You can, but it is not necessary.
In my mind, the view owns its subviews, not the view controller. If one of these subviews is removed from its parent view, I don't think the view controller should be retaining it and it doesn't seem like good design that I have to remember to nil the property in the view controller, too. Worse, if I remove a container view that has nested subviews, I don't want to have to manually keep track of setting all of those individual properties in the view controller that I have to nil, as well.
Don't get me wrong: You can use strong with the subviews if you want/need. But I think it's incorrect to imply that it's ill-advised to use weak.
As a simple rule of thumb,
Any view is always owned (maintained by strong reference) by its
superview, when they are added to the superview.
A top-level view is always owned (maintained by strong reference) by its VC.
Now, a view can be added to its superview in two different ways,
When a view is created from code and added to some superview using addSubview: method. (This is what your question suggest).
When the view hierarchy is loaded from a nib file. (Here also implicit addSubview: calls are made by UIKit to create the hierarchy) and thus subviews are retained by strong reference their respective superviews.
So in either cases views are implicitly retained by their superviews and thus never get disposed until/unless the top-level view gets disposed by itself. So declaring strong properties for subviews imposes another ownership on the subviews:
An implicit ownership by its superview, and
An explicitly ownership by the property.
So in general, declaring strong property for subviews are not required, however it has been created (from nib / by code).
However, under some special situation declaring strong properties for views might be required. For example, when a view should be removed and re-added to its superview, the view should be declared as strong. When such a view is removed from its superview, the superview releases its ownership as well. So if we need to maintain a strong reference of that view if we need to reassign the same view again to some superview. In that case, a strong property reference to the view becomes handy and it disallows the view to be released.
PS:
When creating views with storyboards or nibs I would connect my
outlets. I understand that the outlets are creating a strong
connection to the properties on the view.
I found that there's merely a misconception, in your question. When creating view from storyboard/nib, outlets does not create any so-called strong connection with the properties. IBOutlets only describe how a property loads a view. In case of an outlet property, the view pointed by the property is unarchived and loaded if it has not already been loaded. However, its retention still depends on the strong/weak property accessor.
After reading some tutorials online, I've seen 2 different ways to declare properties.
I'm wondering if one is more correct than the other:
Option 1:
#import <UIKit/UIKit.h>
#interface AppDelegate : NSObject
#property (strong, nonatomic) UIWindow *window;
#property (strong, nonatomic) UITabBarController *mytabs;
#end
Option 2:
#import <UIKit/UIKit.h>
#interface AppDelegate : NSObject {
UIWindow *window;
UITabBarController *mytabs;
}
#property (nonatomic, retain) UIWindow *window;
#property (nonatomic, retain) UITabBarController *mytabs;
#end
The first option uses Automated Reference Counting (ARC); the second one does not. Neither one is "more correct" than the other - the first one is simply more modern. You should use ARC unless you have strong reasons not to (e.g. because of the need to support legacy code on legacy platforms).
Here is a link to learn more about ARC.
Option 1 has ARC enabled. Since iOS5 SDK it's the default option in Xcode.
strong means that the view controller will manage this variable
Here is a good ARC tutorial for iOS5.
Option 2 has Automatic Reference Counting disabled.
the view controller will retain (increase the retain count) when the variable is set
It used to be necessary to explicitly declare an instance variable to back each property. With the "modern" runtime, that's no longer required -- your #synthesize directive for the property will take care of creating the iVar if one doesn't already exist. Some people still keep the ivar declaration, either out of preference or just habit, and of course you'll see the ivars declared in older code. One benefit of doing it is that you can more easily see the values of your properties in the debugger.
The difference between using strong and retain is a matter of using ARC or not, respectively.