I have been running into an issue in using StackMob as the backend of my iOS application (though I'm not sure if this is an issue in wrongly using StackMob's methods or an iOS issue).
I am allowing a user to create a post object that is just a subclassed NSManagedObject, and uploading that to the server to be used in other parts of the application. The issue that arises occurs in the method:
[NSManagedObjectContext saveOnSuccess:<^(void)successBlock> onFailure:<^(NSError *error)failureBlock>];
Here, I am using a StackMob method for asynchronously saving the MOC found in the NSManagedObjectContext(Concurrency) Category Reference.
The view before this one performs a fetch on recent posts, and in the case where the fetch is not performed posting works fine, but if a fetch was performed then in saving the MOC in order to upload the new post I receive the following output as an error message:
2013-09-11 17:08:09.284 imageTagging[1824:1843] -[__NSDictionaryI bytes]: unrecognized
selector sent to instance 0x1e3123d0
2013-09-11 17:08:09.291 imageTagging[1824:1843] *** Terminating app due to uncaught
exception 'NSInvalidArgumentException', reason: '-[__NSDictionaryI bytes]: unrecognized
selector sent to instance 0x1e3123d0'
*** First throw call stack:
(0x318cb3e7 0x395c6963 0x318cef31 0x318cd64d 0x31825208 0x321631cf 0x3216b991 0x15ea99
0x318c8757 0x15e109 0x15dabf 0x10d1c3 0x318d05b7 0x10cd4d 0x10c829 0x10923b 0x1076d9
0x3166c431 0x316c44d1 0x1685c3 0x316c7e5d 0x399e3b3b 0x399e167d 0x399e4613 0x399e47d9
0x39a087f1 0x39a08684)
libc++abi.dylib: terminate called throwing an exception
(lldb)
The data is still uploading to the StackMob server, and can be called upon when the app is run later -- but the app crashes in trying to save it. All of this is performed in the view controller. I've tried to enforce all MOC saves to be performed on the main thread, but the error still occurs. I've also tried dispatching a "save queue" and updating the UI after save completes. This method seemed to work for a bit, but then the errors came up again (may have just been a fluke). I also tried to do this with the synchronous save calls in the documentation
The same error occurs when trying to perform other saves as well (such as after creating a new user or when updating a user's information), and all come down to the same function call causing the problems. It may also be worthwhile to note that the error is always the same (specifically that a type __NSDictionaryI is trying to access its unrecognized selector bytes.
Here is the full method call with the input parameters filled out:
//save context
[[[[SMClient defaultClient] coreDataStore] contextForCurrentThread] saveOnSuccess:^{
NSLog(#"You created a new Post object!");
[[[[SMClient defaultClient] coreDataStore] contextForCurrentThread] refreshObject:newPost mergeChanges:YES];
NSLog(#"refreshed");
} onFailure:^(NSError *error) {
NSLog(#"There was an error! %#", error);
}];
UPDATE: I have narrowed down the problem to a mishandling of information returned from a fetch performed by the previous view controller. Specifically, it occurs after the results are fetched in trying to use the data to update.
As a result of this new insight, the question I am really facing is how to properly save a managed object in the context after a fetch. I believe StackMob takes care of creating the managed objects after the fetch (i.e. server query). I've tried creating a new object from the results array (each "obj" is an NSManagedObject) with:
[results enumerateObjectsUsingBlock:^(id obj, NSUInteger idx, BOOL *stop) {
NSManagedObject *newObj = obj;
}];
I've also tried referencing the fetched results by object id (each "obj" is an objectID) with:
[results enumerateObjectsUsingBlock:^(id obj, NSUInteger idx, BOOL *stop) {
NSManagedObject *newObj = [self.managedObjectContext objectWithID:obj];
}];
Any insight on how to properly do this would be greatly appreciated!
UPDATE 2: It looks like the error is actually occurring in trying to use and save geolocation data. In order to use the queried objects' geolocation data, it must be unarchived -- but to save it, it must be archived. I'm looking into how to do this now, and if I come across a good solution, I'll update again.
FINAL UPDATE: Got it figured out! It turns out that the issue I was having was that I was unarchiving the geolocation data to update the UI and do some calculations, and while I was archiving it again to be stored properly, I created an annotation on a map that referenced the unarchived data. As a result, the MOC maintained the data that could not be saved via the StackMob methods. By only saving the archived data, I can save as often as I'd like and just unarchive the geodata when it needs to be used. Problem solved!
Please feel free to comment if anyone comes across a similar problem and needs some insight or references!
I am just going to put my final update in here as an answer, since it explains how I solved the issue.
FINAL UPDATE: Got it figured out! It turns out that the issue I was having was that I was unarchiving the geolocation data to update the UI and do some calculations, and while I was archiving it again to be stored properly, I created an annotation on a map that referenced the unarchived data. As a result, the MOC maintained the data that could not be saved via the StackMob methods. By only saving the archived data, I can save as often as I'd like and just unarchive the geodata when it needs to be used. Problem solved!
Please feel free to comment if anyone comes across a similar problem and needs some insight or references!
Moral of the story, if you are running into issues similar to this, make sure you are not (even if you don't mean to be) storing references to the unarchived SMGeoPoint data in any of your managed objects. It is trying to store those that causes the problem.
Related
iOS Core Data - Serious application error - attempt to insert nil
Hello,
My app runs actualy stable, but in seldom cases it crashes with this error message...
2019-04-02 20:48:52.437172+0200 myAppName[4422:1595677] [error] error: Serious application error. Exception was caught during Core Data change processing. This is usually a bug within an observer of NSManagedObjectContextObjectsDidChangeNotification. -[__NSCFSet addObject:]: attempt to insert nil with userInfo (null)
CoreData: error: Serious application error. Exception was caught during Core Data change processing. This is usually a bug within an observer of NSManagedObjectContextObjectsDidChangeNotification. -[__NSCFSet addObject:]: attempt to insert nil with userInfo (null)
2019-04-02 20:48:52.438246+0200 myAppName[4422:1595677] *** Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: '-[__NSCFSet addObject:]: attempt to insert nil'
...when it tries to save the current context (this part in my code is still in objc):
- (void)saveChanges
{
dispatch_async(dispatch_get_main_queue(), ^{
NSError *err = nil;
BOOL succesful = [self->context save:&err];
if (!succesful)
{
NSLog(#"ERROR MESSAGE DURING SAVING CONTEXT: %#", [err localizedDescription]);
}
});
}
'seldom' means:
Most Customers do never experience the issue, for few customers it happens several times per day.
I was able to produce it 2 times during the last two days although I tried several ways to force this error (see below).
This is the setup:
The respective data is in one Entity (table)
A NSFetchedResultsController shows the data in an UITableView
User can hit a button to add a new record.
New record has only some basic data and initiates two API calls to two webservers
Each webserver response does update the record
After both are done (or were cancelled due to timeout), I call the saveChanges function from above only once.
All functions use the same context created by NSPersistentContainer as follow (this part is already in swift)
#objc lazy var persistentContainer: NSPersistentContainer = {
let container = NSPersistentContainer(name: "myAppName")
let description = NSPersistentStoreDescription(url: SomeHelper.urlForFileInDocFolder("storev7.data"))
container.persistentStoreDescriptions = [description]
container.loadPersistentStores(completionHandler: { (storeDescription, error) in
if let error = error as NSError? {
fatalError("Unresolved error \(error), \(error.userInfo)")
}
})
return container
}()
If I could reproduce the error somehow, I could find an appropriate solution, but as it almost never happens, I'm stuck.
Do you have an idea how I could reproduce the error from above? Or do you have a clue what could cause the error in my case?
What I tried already to reproduce the error:
Create hundereds of record
Create hundereds of record in a few seconds
Create hundereds of record during switching internet connection on / off / on / off /...
Create hundereds of record during mixed from background and main thread (I removed the dispatch from saveChanges for that)
Create hundereds of record with different delays on the API (added random sleep function on the webserver)
Long time execution, the app run for 24 hours on a real device and created record each 2 minutes
Mixes of all of them
NSManagedObjects are restricted to a single queue. They are not thread-safe for reading or writing. Reading an NSManagedObject can cause a fault, which is a write operation. That means that NSManagedObjects retrieved from a main queue context (like viewContext) cannot be passed to other queues.
The details of all of this are discussed in the Core Data Programming Guide:
NSManagedObject instances are not intended to be passed between queues. Doing so can result in corruption of the data and termination of the application. When it is necessary to hand off a managed object reference from one queue to another, it must be done through NSManagedObjectID instances.
The general approach with NSPersistentContainer is to use something like viewContext exclusively on the main queue, and to use performBackgroundTask to handle background operations, or you can use newBackgroundContext to generate a background context, and use perform or performAndWait on it to manipulate objects that are fetched from that context.
Moving an object between contexts is done by fetching the same objectID in the other context (keeping in mind that this will return a fresh instance from the store).
You can track down mistakes by adding -com.apple.CoreData.ConcurrencyDebug 1 to your scheme. When you do this, errors will immediately trap on the delightfully named __Multithreading_Violation_AllThatIsLeftToUsIsHonor__.
I have a problem/crash merging the data of different NSManagedObjectContexts (iOS 6.1, Xcode 4.6).
Most of the time the error that rises is the following:
CoreData: error: Serious application error. Exception was caught during Core Data change processing. This is usually a bug within an observer of NSManagedObjectContextObjectsDidChangeNotification. statement is still active with userInfo (null)
One time i got this error:
An observer of NSManagedObjectContextDidSaveNotification illegally threw an exception. Objects saved = { $OBJLIST } and exception = statement is still active with userInfo = (null)
Sadly there is no value in the stacktraces i got. They just show symbols that are CoreData internal (if any).
Our CoreData stack:
1 NSPersistentStoreLocator shared by all threads
1 unique MOC per thread ( created on first need )
All MOCs are saved in a Dictionary
An observer is added for the notification NSManagedObjectContextDidSaveNotification to update the MOCs when one is saving to the store. The defined selector is calling mergeChangesFromContextDidSaveNotification on every other thread/context except the one that did the save operation.
+ (void)mergeChanges:(NSNotification *)notification {
NSManagedObjectContext *ctx;
for ( NSNumber *threadId in [__managedObjectContexts keyEnumerator] ) {
ctx = [__managedObjectContexts objectForKey:threadId];
if ( notification.object != ctx ) {
[ctx mergeChangesFromContextDidSaveNotification:notification];
}
}
}
Steps to produce the error:
In a background thread CoreData data objects that arent needed anymore (unreferenced by other objects) are being deleted.
[[CDUtils managedObjectContext] deleteObject:obj];
[[CDUtils managedObjectContext] save:&error];
While this is happening the user can navigate throughout the application. Userinteraction (i.e. opening a tableview) can trigger executeFetch calls on the moc of the main thread.
Every thread uses the same NSPersistentStoreLocator but a different/unique MOC.
We tried different methods of locking with NSLocks and the lock on the NSPersistentStoreLocator for threadsafety. I.e. enclosing the mergeChanges Method and the save operation each by a lock/unlock or enclosing both methods in the same lock/unlock. Sadly we had no succes thus far.
[__storeCoordinator lock];
[__storeCoordinator unlock];
I'd be thankful for every piece of advice you can give me to approach a solution. Thank you for your time!
For the people interested. I managed to make things work with multiple threads / MOCs. I basicly solved the original problem / those errors i had by locking PSC and MOCs correctly. The next problem that arose was how to know if its save to mergeChanges on a context. I cant lock nor should mergeChanges on a MOC that has no running thread anymore. But how do i know if the thread is running or not? If i just check for NSThreads "isExecuting"-Method it might happen that the thread exits just after i checked the BOOL. Im trying an easier approach now where i just merge into the main thread.
My iOS app uses core data via multiple threads. I am getting some crash reports with the following message: "'NSObjectInaccessibleException', reason: 'CoreData could not fulfill a fault for '0x1e07a9b0 ''
I understand what is causing this problem - that the object was deleted but another thread is trying to access it. I am working to solve the problem but I want to add a check in the background thread to see if the object will fault in this manner.
My code at the moment relates to myObject.myValue. Is it possible to do some check, such as:
if (!myObject.myValue) {
return;
}
... so that it will get out of the method before doing anything that could cause such a crash? Or will simply calling myObject.myValue, even to see if it's null, cause such an exception to be thrown?
You could try and use existingObjectWithID:error::
Returns the object for the specified ID.
- (NSManagedObject *)existingObjectWithID:(NSManagedObjectID *)objectID error:(NSError **)error
Discussion
If there is a managed object with the given ID already registered in the context, that object is returned directly; otherwise the corresponding object is faulted into the context.
This method might perform I/O if the data is uncached.
Unlike objectWithID:, this method never returns a fault.
You could dO:
if ([myMOC existingObjectWithID:myObject.objectID error:&error])
...
You should verify that the object exists before accessing it's variables if you're having issues where the object may be deleted on another thread.
Two methods:
Refresh the view datasources whenever your data is being deleted. You can do this by registering for the NSManagedObjectContextObjectsDidChangeNotification notification and then parsing the userInfo on that notification to see which object was deleted.
Use code similar to below when you're passing data around to multiple threads.
Example:
// Cache and pass the object's ID off to another thread to do work on
// You can just store it as a property on the class
#try {
NSManagedObject *theObject = [managedObjectContext objectWithID:self.theObjectID];
// do stuff with object
}
#catch (NSException * e) {
// An entity with that object ID could not be found (maybe they were deleted)
NSLog(#"Error finding object: %#: %#", [e name], [e reason]);
}
You can check the NSManagedContext is existed when you use the NSManagedObject.
like this:
if (obj.managedObjectContext)
{
//do things
}
You can check [myObject isFault] where myObject is a NSManagedObject instance
You could give a try to use :
shouldDeleteInaccessibleFaults
property on managed object context. As this article says it should change the behaviour of faulting already deleted object.
https://cocoacasts.com/what-are-core-data-query-generations/
Edit:
Since iOS 9 (when it was added) this property default value is YES.
The good facts:
I download data from the server, and, via Core Data thread confinement, save the data, and when the background MOC is saved, the main MOC gets merged.
All the saving operations go ok
Also the merging of the MOC happens without any problems
The bug I'm hunting:
When my UITableView with NSFetchedResultsController is active (i.e. on the screen), and the saving is happening, the app crashes with a SIGABRT that takes me to the mergeChangesFromContextDidSaveNotification line in AppDelegate.
What is the most strange part is, that when the delegate of the FRC is nil, or when it is my view controller but i don't implement any FRC delegate methods, the crash doesn't happen and I don't have any problem. But when I implement any of the delegate methods (even empty, without a single line of code), the app crashes with the same bug. It means that the methods are not even being fired, the problem is not in the code inside these methods.
The strangest part 2 (CHECK UPDATE 2 BELOW): the crash happens with a [__NSCFNumber length]: unrecognized selector sent to instance and I don't call any 'length' property in my CoreDataManager neither in my AppDelegate class
The witness: console
<CoreDataManager.m:(338)> Saved data from server
<AppDelegate.m:(352)> Will merge
<CoreDataManager.m:(338)> Saved data from server
<AppDelegate.m:(355)> Did merge
<CoreDataManager.m:(338)> Saved data from server
<AppDelegate.m:(352)> Will merge
<AppDelegate.m:(355)> Did merge
<CoreDataManager.m:(338)> Saved data from server
<CoreDataManager.m:(338)> Saved data from server
<CoreDataManager.m:(338)> Saved data from server
<MyTableViewController.m:(134)> Fetched results controller did fetch
<CoreDataManager.m:(338)> Saved data from server
<CoreDataManager.m:(338)> Saved data from server
<CoreDataManager.m:(338)> Saved data from server
<AppDelegate.m:(352)> Will merge
<CoreDataManager.m:(338)> Saved data from server
[__NSCFNumber length]: unrecognized selector sent to instance 0x13318050
Some code - Merging the MOCs
- (void)managedObjectContextDidSave:(NSNotification *)notification
{
NSManagedObjectContext *sender = (NSManagedObjectContext *)[notification object];
if ((sender != self.managedObjectContext) &&
(sender.persistentStoreCoordinator == self.managedObjectContext.persistentStoreCoordinator))
{
dispatch_async(dispatch_get_main_queue(), ^{
DebugLog(#"Will merge");
[self.managedObjectContext mergeChangesFromContextDidSaveNotification:notification];
DebugLog(#"Did merge");
});
}
}
Update 1
Following Cocoanetics hint, I created a NSNumber category to check who is calling length. I got what you see below, and a crash in [__NSCFNumber _fastCStringContents:]: unrecognized selector sent to instance.
Update 2
Enabling zombies didn't help =(
Make sure that you are only observing the notification from other MOCs. If you save there this triggers another such notification and you might be going in an endless loop that fails after one or two iterations because an object had been released by ARC.
Yours sounds like a memory issue. Check your ARC ownership qualifiers and enable NSZombies. Enabling NSZombies will help you narrow down the object that was released prematurely.
When you enable zombies you will see a "message sent to a deallocated instance" instead. Check which object was released prematurely and update your question.
Well, after months and hours, I finally got a solution. It works, and I would love to hear some opinions on why.
So, as I said, the saving was working 100%, as well as the merging notifications. If I set the NSFetchedResultsController delegate to nil, there was no problem. However, setting the delegate to my UIViewController, made the app crash.
I thought it could be, maybe due to my code when the delegate methods were triggered. But the app crashed even before that. So I followed Cocoanetics tip, to create a category and try to figure out who was calling the length method to the NSNumber object. After that, I saw that the NSPredicate was calling - (BOOL)evaluateWithObject:(id)object; before getting to the crash. In the same way, I did a category to override it:
#interface NSPredicate (PractiPredicate)
- (BOOL)evaluateWithObject:(id)object;
#end
#implementation NSPredicate (PractiPredicate)
- (BOOL)evaluateWithObject:(id)object
{
NSLog(#"Evaluate was called. Object class %#", NSStringFromClass([object class]));
MyManagedObject *myManagedObject = object;
NSLog(#"Is fault? %d", myManagedObject.isFault);
NSLog(#"myManagedObject changed and already have propertyA? %d", myManagedObject.propertyA != nil);
return YES;
}
#end
So, for my surprise, it worked, and generated te following logs:
Evaluate was called. Object class MyManagedObject
Is fault? 0
myManagedObject changed and already have propertyA? 1
I decided to print "Is fault?" because I thought that this mess was related to NSManagedObject faulting, but, for what it printed, it's not.
Question for the comments: What do you think that could have generated this problem here?
Most likely the problem does not lie in the code you posted but in how you deal with the changes in the fetched results controller delegate. Those are just being triggered by the merge.
I had the same problem and in my case, i found out that the reason for the crash was an incorrect predicate. I had a predicate like this:
NSPredicate *predicate = [NSPredicate predicateWithFormat:#"attribute > 0"];
Where attribute was a string. I corrected it as:
NSPredicate *predicate = [NSPredicate predicateWithFormat:#"attribute.length > 0"];
Now, my code is running fine. Make sure you check all the predicates in your code since this can also be a reason for this crash.
I have researched a ton of posts regarding Core Data on background threads, and I feel like I understand (on paper) what needs to be going on. I guess we'll see. I am working on migrating an existing OS X app to Core Data, and am having issues making new instances of my NSManagedObject on an async thread.
Here is a sample of the code I am running right after I have moved onto a background thread:
NSLog(#"JSON 1");
NSManagedObjectContext * context = [[NSManagedObjectContext alloc] init];
[context setPersistentStoreCoordinator:[[NSApp delegate] persistentStoreCoordinator]];
asset = (MTAssetInfo*)[NSEntityDescription insertNewObjectForEntityForName:#"Info" inManagedObjectContext:context];
NSLog(#"JSON 2");
The result is that the first log message (#"JSON 1") gets called 31 times, and the second one (#"JSON 2") is never called. The object isn't being made and returned correctly.
The model for this Info entity is quite complex with a few transformable attributes that may or may not be setup correctly. The weird thing is that similar code run on the main thread and the main MOC works great. No issues.
EDIT - Some more context
The async call originates from here:
for (NSNumber *sectionID in sectionsToShow) {
dispatch_group_async(group, queue, ^{
MTAssetInfo *asset = [self assetWithRefID:[sectionID unsignedIntegerValue]];
if (asset != nil) {
[sectionsLock lock];
[sectionsTemp addObject:asset];
[sectionsLock unlock];
}
});
}
The assetWithRefID method never returns with an object because of the other code snippet. It never successfully pulls an NSManagedObject out of the context on the background thread.
You are going to have to provide more information to get real help, but I bet your problem is an error happening in the NSManagedDocument background thread.
I'd register a NSNotificationCenter for ALL messages (name:nil object:nil) and just print them out. I bet you see a status change or error message in there that is failing.
You might want to try a #try/#catch block around it just to see if exceptions are being thrown.
Maybe it will give you more to go on.
One other suggestion... Swizzling isn't necessarily the right tool for production stuff, but it's almost unbeatable for debugging. I have method-swizled several entire classes, so that it sends a detailed NSNotification before/after each invocation.
It has saved me tons of time, and helped me track down some wicked bugs. Now, when something is going on in CoreData, I take out my set of classes, link them in, and see all the detail I want.
I know that does not exactly answer you question, but hopefully it will put you on the track so you can provide some more information and get it all fixed.
If that's too much for you, create a subclass and instantiate that, with a similar method for calling super. You can get a real idea of the entire flow pretty easily.