The following code is from the LazyTableImages sample code provided by Apple (source here).
In their completion block they have a reference to self which should cause a retain cycle... But I don't get a warning for this in Xcode whereas in similar code of mine I would.
Is this correct?
Perhaps I'm missing a subtlety of this.
- (void)startIconDownload:(AppRecord *)appRecord forIndexPath:(NSIndexPath *)indexPath
{
IconDownloader *iconDownloader = [self.imageDownloadsInProgress objectForKey:indexPath];
if (iconDownloader == nil)
{
iconDownloader = [[IconDownloader alloc] init];
iconDownloader.appRecord = appRecord;
[iconDownloader setCompletionHandler:^{
UITableViewCell *cell = [self.tableView cellForRowAtIndexPath:indexPath];
// Display the newly loaded image
cell.imageView.image = appRecord.appIcon;
// Remove the IconDownloader from the in progress list.
// This will result in it being deallocated.
[self.imageDownloadsInProgress removeObjectForKey:indexPath];
}];
[self.imageDownloadsInProgress setObject:iconDownloader forKey:indexPath];
[iconDownloader startDownload];
}
}
The retain cycle that you think you are seeing is because the object holds the the downloader in a dictionary.
It's true that there is a strong reference to self in the block, but, as long as the completion handler is always run, the downloader will be removed from the dictionary. And eventually this dictionary will be empty, which means there will be no objects holding on to self, and thus no retain cycle.
self doesn't have a strong pointer to iconDownloader. It's created and scoped just to this method:
IconDownloader *iconDownloader = [self.imageDownloadsInProgress objectForKey:indexPath];
If iconDownloader was a strong property (self.iconDownloader) then Xcode would detect a strong reference cycle.
Capturing self itself is no retain cycle. It is a single reference. One reference cannot build a cycle. The usual antipattern is, that additionale a reference to the block is stored in a strong property of self. Than there are two references building a cycle.
There's no warning because the compiler isn't yet capable of detecting all possible retain cycles.
For example:
- (void)foo
{
_block = ^ { [self done]; }; // Warning: Possible retain cycle
DSGenericBlock foo = ^ { [self done] };
_block = foo; // No warning.
}
If you were to assign the block directly to an instance variable of "self", you would get the "possible retain cycle" warning. Instead, the block is assigned to another object, which is then retained by self, so the compiler does not detect the cycle (even though the cycle does exist).
Related
I have a "always keep a static reference on the last created object" pattern for one of my classes.
When I create a new instance of it, the static pointer should been set to this new instance.
Solved the way below:
static AViewController *actualSelf = nil;
+ (AViewController *) getActualSelf {
return actualSelf;
}
- (AViewController *) init {
self = [super init];
self.title = #"Title";
actualSelf = nil; // Why do I need this line the second time init gets called?
actualSelf = self; // this is the only line which sets actualSelf
return self;
}
- (void)dealloc {
actualSelf = nil;
}
…SOME IMPORTANT MUST BE MISSING, BUT I DON'T KNOW WHAT
create AViewController a > Init gets called > Static pointer is set to a
create AViewController b > Init gets called > Static pointer should be set to b. But the static pointer is not set to b! It's sets itself from a to nil!
Object a is most of the time deallocated before b gets created.
Do I always need to reset the static variables to nil when deallocating them? NO, see my own answer below.
Without the line "actualSelf = nil;", through ARC object a gets deallocated in the moment actualSelf is set to object b.
actualSelf = nil; // WHY DO I NEED THIS LINE?
You don't need the line, it accomplishes nothing and will probably be eliminated by the compiler optimizer in a release build.
The class is not a singleton.
The rest of the answer has been deleted because the question has been modified.
create AViewController a > Init gets called > Static pointer is set to a
create AViewController b > Init gets called > Static pointer should be set to b. But the static pointer is not set to b! It's sets itself
from a to nil!
Probably nothing other than actualSelf has ownership of a. Therefore, actualSelf = self; during the initialization of b causes a to be fully released and thus deallocated. During the -dealloc of a, it sets actualSelf to nil. Apparently, this is coming after the assignment of actualSelf to the new self (object b). (The precise order of operations of the ARC-supplied -release call vis-a-vis the assignment of actualSelf is unpredictable.)
The very fact that you are assigning actualSelf to nil in the -dealloc method smells bad. It suggests that actualSelf should be __weak because if it's holding a strong reference to an object, then that object can't be deallocated.
At the very least, -dealloc should test if actualSelf equals self before resetting it. But that will never be true so long as actualSelf is a strong reference, since a strongly referenced object can't be deallocated.
This whole approach is deeply problematic. What are you really trying to accomplish?
How can I avoid this warning in xcode. Here is the code snippet:
[player(AVPlayer object) addPeriodicTimeObserverForInterval:CMTimeMakeWithSeconds(0.1, 100)
queue:nil usingBlock:^(CMTime time) {
current+=1;
if(current==60)
{
min+=(current/60);
current = 0;
}
[timerDisp(UILabel) setText:[NSString stringWithFormat:#"%02d:%02d",min,current]];///warning occurs in this line
}];
The capture of self here is coming in with your implicit property access of self.timerDisp - you can't refer to self or properties on self from within a block that will be strongly retained by self.
You can get around this by creating a weak reference to self before accessing timerDisp inside your block:
__weak typeof(self) weakSelf = self;
[player addPeriodicTimeObserverForInterval:CMTimeMakeWithSeconds(0.1, 100)
queue:nil
usingBlock:^(CMTime time) {
current+=1;
if(current==60)
{
min+=(current/60);
current = 0;
}
[weakSelf.timerDisp setText:[NSString stringWithFormat:#"%02d:%02d",min,current]];
}];
__weak MyClass *self_ = self; // that's enough
self.loadingDidFinishHandler = ^(NSArray *receivedItems, NSError *error){
if (!error) {
[self_ showAlertWithError:error];
} else {
self_.items = [NSArray arrayWithArray:receivedItems];
[self_.tableView reloadData];
}
};
And one very important thing to remember:
do not use instance variables directly in block, use it as a properties of weak object, sample:
self.loadingDidFinishHandler = ^(NSArray *receivedItems, NSError *error){
if (!error) {
[self_ showAlertWithError:error];
} else {
self_.items = [NSArray arrayWithArray:receivedItems];
[_tableView reloadData]; // BAD! IT ALSO WILL BRING YOU TO RETAIN LOOP
}
};
and don't forget to do:
- (void)dealloc {
self.loadingCompletionHandler = NULL;
}
another issue can appear if you will pass weak copy of not retained by anybody object:
MyViewController *vcToGo = [[MyViewCOntroller alloc] init];
__weak MyViewController *vcToGo_ = vcToGo;
self.loadingCompletion = ^{
[vcToGo_ doSomePrecessing];
};
if vcToGo will be deallocated and then this block fired I believe you will get crash with unrecognized selector to a trash which is contains vcToGo_ variable now. Try to control it.
Better version
__strong typeof(self) strongSelf = weakSelf;
Create a strong reference to that weak version as the first line in your block. If self still exists when the block starts to execute and hasn’t fallen back to nil, this line ensures it persists throughout the block’s execution lifetime.
So the whole thing would be like this:
// Establish the weak self reference
__weak typeof(self) weakSelf = self;
[player addPeriodicTimeObserverForInterval:CMTimeMakeWithSeconds(0.1, 100)
queue:nil
usingBlock:^(CMTime time) {
// Establish the strong self reference
__strong typeof(self) strongSelf = weakSelf;
if (strongSelf) {
[strongSelf.timerDisp setText:[NSString stringWithFormat:#"%02d:%02d",min,current]];
} else {
// self doesn't exist
}
}];
I have read this article many times. This is an excellent article by Erica Sadun on
How To Avoid Issues When Using Blocks And NSNotificationCenter
Swift update:
For example, in swift a simple method with success block would be:
func doSomeThingWithSuccessBlock(success: () -> ()) {
success()
}
When we call this method and need to use self in the success block. We'll be using the [weak self] and guard let features.
doSomeThingWithSuccessBlock { [weak self] () -> () in
guard let strongSelf = self else { return }
strongSelf.gridCollectionView.reloadData()
}
This so-called strong-weak dance is used by popular open source project Alamofire.
For more info check out swift-style-guide
In another answer, Tim said:
you can't refer to self or properties on self from within a block that will be strongly retained by self.
This isn’t quite true. It’s OK for you to do this so long as you break the cycle at some point. For example, let’s say you have a timer that fires that has a block that retains self and you also keep a strong reference to the timer in self. This is perfectly fine if you always know that you will destroy the timer at some point and break the cycle.
In my case just now, I had this warning for code that did:
[x setY:^{ [x doSomething]; }];
Now I happen to know that clang will only produce this warning if it detects the method starts with “set” (and one other special case that I won’t mention here). For me, I know there is no danger of there being a retain loop, so I changed the method name to “useY:” Of course, that might not be appropriate in all cases and usually you will want to use a weak reference, but I thought it worth noting my solution in case it helps others.
Many times, this is not actually a retain cycle.
If you know that it's not, you need not bring fruitless weakSelves into the world.
Apple even forces these warnings upon us with the API to their UIPageViewController, which includes a set method (which triggers these warnings–as mentioned elsewhere–thinking you are setting a value to an ivar that is a block) and a completion handler block (in which you'll undoubtedly refer to yourself).
Here's some compiler directives to remove the warning from that one line of code:
#pragma GCC diagnostic push
#pragma clang diagnostic ignored "-Warc-retain-cycles"
[self.pageViewController setViewControllers:#[newViewController] direction:navigationDirection animated:YES completion:^(BOOL finished) {
// this warning is caused because "setViewControllers" starts with "set…", it's not a problem
[self doTheThingsIGottaDo:finished touchThePuppetHead:YES];
}];
#pragma GCC diagnostic pop
Adding two cents on improving precision and style. In most cases you will only use one or a couple of members of self in this block, most likely just to update a slider. Casting self is overkill. Instead, it's better to be explicit and cast only the objects that you truly need inside the block. For example, if it's an instance of UISlider*, say, _timeSlider, just do the following before the block declaration:
UISlider* __weak slider = _timeSlider;
Then just use slider inside the block. Technically this is more precise as it narrows down the potential retain cycle to only the object that you need, not all the objects inside self.
Full example:
UISlider* __weak slider = _timeSlider;
[_embeddedPlayer addPeriodicTimeObserverForInterval:CMTimeMake(1, 1)
queue:nil
usingBlock:^(CMTime time){
slider.value = time.value/time.timescale;
}
];
Additionally, most likely the object being cast to a weak pointer is already a weak pointer inside self as well minimizing or eliminating completely the likelihood of a retain cycle. In the example above, _timeSlider is actually a property stored as a weak reference, e.g:
#property (nonatomic, weak) IBOutlet UISlider* timeSlider;
In terms of coding style, as with C and C++, variable declarations are better read from right to left. Declaring SomeType* __weak variable in this order reads more naturally from right to left as: variable is a weak pointer to SomeType.
I ran into this warning recently and wanted to understand it a bit better. After a bit of trial and error, I discovered that it originates from having a method start with either "add" or "save". Objective C treats method names starting with "new", "alloc", etc as returning a retained object but doesn't mention (that I can find) anything about "add" or "save". However, if I use a method name in this way:
[self addItemWithCompletionBlock:^(NSError *error) {
[self done]; }];
I will see the warning at the [self done] line. However, this will not:
[self itemWithCompletionBlock:^(NSError *error) {
[self done]; }];
I will go ahead and use the "__weak __typeof(self) weakSelf = self" way to reference my object but really don't like having to do so since it will confuse a future me and/or other dev. Of course, I could also not use "add" (or "save") but that's worse since it takes away the meaning of the method.
I've seen several other questions of the same form, but I either a) can't understand the provided answers, or b) don't see how those situations are similar to mine.
I'm writing a Category on UIView to recursively evaluate all the subviews of a UIView and return an Array of subviews passing a test. I've noted where my compiler warning occurs:
-(NSArray*)subviewsPassingTest:(BOOL(^)(UIView *view, BOOL *stop))test {
__block BOOL *stop = NO;
NSArray*(^__block evaluateAndRecurse)(UIView*);
evaluateAndRecurse = ^NSArray*(UIView *view) {
NSMutableArray *myPassedChildren = [[NSMutableArray alloc] init];
for (UIView *subview in [view subviews]) {
BOOL passes = test(subview, stop);
if (passes) [myPassedChildren addObject:subview];
if (stop) return myPassedChildren;
[myPassedChildren addObjectsFromArray:evaluateAndRecurse(subview)];
// ^^^^ Compiler warning here ^^^^^
// "Capturing 'evaluateAndRecurse' strongly in this block
// is likely to lead to a retrain cycle"
}
return myPassedChildren;
};
return evaluateAndRecurse(self);
}
Also, I get a bad_access failure when I don't include the __block modifier in my block's declaration (^__block evaluateAndRecurse). If someone could explain why that is, that would be very helpful too. Thanks!
The problem here is that your block evaluteAndRecurse() captures itself, which means that, if it's ever to be copied (I don't believe it will in your case, but in slightly less-trivial cases it may), then it will retain itself and therefore live forever, as there is nothing to break the retain cycle.
Edit: Ramy Al Zuhouri made a good point, using __unsafe_unretained on the only reference to the block is dangerous. As long as the block remains on the stack, this will work, but if the block needs to be copied (e.g. it needs to escape to a parent scope), then the __unsafe_unretained will cause it to be deallocated. The following paragraph has been updated with the recommended approach:
What you probably want to do here is use a separate variable marked with __unsafe_unretained that also contains the block, and capture that separate variable. This will prevent it from retaining itself. You could use __weak, but since you know that the block must be alive if it's being called, there's no need to bother with the (very slight) overhead of a weak reference. This will make your code look like
NSArray*(^__block __unsafe_unretained capturedEvaluteAndRecurse)(UIView*);
NSArray*(^evaluateAndRecurse)(UIView*) = ^NSArray*(UIView *view) {
...
[myPassedChildren addObjectsFromArray:capturedEvaluateAndRecurse(subview)];
};
capturedEvaluateAndRecurse = evaluteAndRecurse;
Alternatively, you could capture a pointer to the block, which will have the same effect but allow you to grab the pointer before the block instantiation instead of after. This is a personal preference. It also allows you to omit the __block:
NSArray*(^evaluateAndRecurse)(UIView*);
NSArray*(^*evaluteAndRecursePtr)(UIView*) = &evaluateAndRecurse;
evaluateAndRecurse = ^NSArray*(UIView*) {
...
[myPassedChildren addObjectsFromArray:(*evaluateAndRecursePtr)(subview)];
};
As for needing the __block, that's a separate issue. If you don't have __block, then the block instance will actually capture the previous value of the variable. Remember, when a block is created, any captured variables that aren't marked with __block are actually stored as a const copy of their state at the point where the block is instantiated. And since the block is created before it's assigned to the variable, that means it's capturing the state of the capturedEvaluteAndRecurse variable before the assignment, which is going to be nil (under ARC; otherwise, it would be garbage memory).
In essence, you can think of a given block instance as actually being an instance of a hidden class that has an ivar for each captured variable. So with your code, the compiler would basically treat it as something like:
// Note: this isn't an accurate portrayal of what actually happens
PrivateBlockSubclass *block = ^NSArray*(UIView *view){ ... };
block->stop = stop;
block->evaluteAndRecurse = evaluateAndRecurse;
evaluteAndRecurse = block;
Hopefully this makes it clear why it captures the previous value of evaluateAndRecurse instead of the current value.
I've done something similar, but in a different way to cut down on time allocating new arrays, and haven't had any problems. You could try adapting your method to look something like this:
- (void)addSubviewsOfKindOfClass:(id)classObject toArray:(NSMutableArray *)array {
if ([self isKindOfClass:classObject]) {
[array addObject:self];
}
NSArray *subviews = [self subviews];
for (NSView *view in subviews) {
[view addSubviewsOfKindOfClass:classObject toArray:array];
}
}
From the Transitioning to ARC Release Notes
Use Lifetime Qualifiers to Avoid Strong Reference Cycles
You can use lifetime qualifiers to avoid strong reference cycles. For
example, typically if you have a graph of objects arranged in a
parent-child hierarchy and parents need to refer to their children and
vice versa, then you make the parent-to-child relationship strong and
the child-to-parent relationship weak. Other situations may be more
subtle, particularly when they involve block objects.
In manual reference counting mode, __block id x; has the effect of not
retaining x. In ARC mode, __block id x; defaults to retaining x (just
like all other values). To get the manual reference counting mode
behavior under ARC, you could use __unsafe_unretained __block id x;.
As the name __unsafe_unretained implies, however, having a
non-retained variable is dangerous (because it can dangle) and is
therefore discouraged. Two better options are to either use __weak (if
you don’t need to support iOS 4 or OS X v10.6), or set the __block
value to nil to break the retain cycle.
Okay, so what's different about __block variable?
Why set to nil here? Is __block variable retained twice? Who hold all the reference? The block? The heap? The stack? The thread? The what?
The following code fragment illustrates this issue using a pattern that is sometimes used in manual reference counting.
MyViewController *myController = [[MyViewController alloc] init…];
// ...
myController.completionHandler = ^(NSInteger result) {
[myController dismissViewControllerAnimated:YES completion:nil];
};
[self presentViewController:myController animated:YES completion:^{
[myController release];
}];
As described, instead, you can use a __block qualifier and set the myController variable to nil in the completion handler:
MyViewController * __block myController = [[MyViewController alloc] init…]; //Why use __block. my controller is not changed at all
// ...
myController.completionHandler = ^(NSInteger result) {
[myController dismissViewControllerAnimated:YES completion:nil];
myController = nil; //Why set to nil here? Is __block variable retained twice? Who hold all the reference? The block? The heap? The stack? The thread? The what?
};
Also why myController is not set to nil by compiler. Why do we have to do so? It seems that the compiler sort of know when myController will no longer be used again namely when the block expire.
When you have code of this form:
object.block = ^{
// reference object from inside the block
[object someMethodOrProperty];
};
object will retain or copy the block you give to it. But the block itself will also retain object because it is strongly referenced from within the block. This is a retain cycle. Even after the block has finished executing, the reference cycle still exists and neither the object nor the block can be deallocated. Remember that a block can be called multiple times, so it cannot just forget all the variables it references after it has finished executing once.
To break this cycle, you can define object to be a __block variable, which allows you to change its value from inside the block, e.g. changing it to nil to break the cycle:
__block id object = ...;
object.block = ^{
// reference object from inside the block
[object someMethodOrProperty];
object = nil;
// At this point, the block no longer retains object, so the cycle is broken
};
When we assign object to nil at the end of the block, the block will no longer retain object and the retain cycle is broken. This allows both objects to be deallocated.
One concrete example of this is with with NSOperation's completionBlock property. If you use the completionBlock to access an operation's result, you need to break the retain cycle that is created:
__block NSOperation *op = [self operationForProcessingSomeData];
op.completionBlock = ^{
// since we strongly reference op here, a retain cycle is created
[self operationFinishedWithData:op.processedData];
// break the retain cycle!
op = nil;
}
As the documentation describes, there are a number of other techniques you can also use to break these retain cycles. For example, you will need to use a different technique in non-ARC code than you would in ARC code.
I prefer this solution
typeof(self) __weak weakSelf = self;
self.rotationBlock = ^{
typeof (weakSelf) __strong self = weakSelf;
[self yourCodeThatReferenceSelf];
};
What happens is that the block will capture self as a weak reference and there will be no retain cycle. self inside the block is then redefined as __strong self = weakSelf before your code runs. This prevents self from being released while your block runs.
I've got a class method that uses dispatch_once to create a static object. Inside the dispatch_once block I use [self class] and was wondering if I need to use a weak reference to self to avoid a retain cycle?
+ (NSArray *)accountNames{
static NSArray *names = nil;
static dispatch_once_t predicate;
dispatch_once(&predicate, ^{
names = [[[self class] accounts] allKeys];
names = [names sortedArrayUsingSelector:#selector(caseInsensitiveCompare:)];
});
return names;
}
If I use a weak reference to self I get a warning:
+ (NSArray *)accountNames{
static NSArray *names = nil;
static dispatch_once_t predicate;
__weak TUAccount *wself = self;
dispatch_once(&predicate, ^{
names = [[[wself class] accounts] allKeys];
names = [names sortedArrayUsingSelector:#selector(caseInsensitiveCompare:)];
});
return names;
}
Incompatible pointer types initializing 'TUAccount *__weak' with an expression of type 'const Class'
Because I get a warning I don't think I need to use a weak reference to self in this case but I wanted to see what you guys thought.
There is no reason to worry about a retain cycle here, because it's meaningless to retain or release a class object -- retain and release simply have no effect.
Your attempt at making a weak reference is wrong, because you are taking a class object self and trying to cast it to an instance of TUAccount. The two are completely different things.
Also, you can simplify:
names = [[[self class] accounts] allKeys];
Since self is already a class, [self class] == self, so do this instead:
names = [[self accounts] allKeys];
I have checked one more time iOS SDK Docs and found next:
Objective-C Objects
In a manually reference-counted environment, local variables used within the block are retained when the block is copied. Use of instance variables within the block will cause the object itself to be retained. If you wish to override this behavior for a particular object variable, you can mark it with the __block storage type modifier.
If you are using ARC, object variables are retained and released automatically as the block is copied and later released.
Note: In a garbage-collected environment, if you apply both __weak and __block modifiers to a variable, then the block will not ensure that it is kept alive.
If you use a block within the implementation of a method, the rules for memory management of object instance variables are more subtle:
If you access an instance variable by reference, self is retained;
If you access an instance variable by value, the variable is retained.
The following examples illustrate the two different situations:
dispatch_async(queue, ^{
// instanceVariable is used by reference, self is retained
doSomethingWithObject(instanceVariable);
});
id localVariable = instanceVariable;
dispatch_async(queue, ^{
// localVariable is used by value, localVariable is retained (not self)
doSomethingWithObject(localVariable);
});
Conclusion: I assume that there is no problem with using self in block. It will be retained and after execution released.
Moreover, you are not storing block in memory and using it directly. So it is copied to the heap, executed and pushed from it. I don't see any retain cycles.
Hope I am right!