I'm experiencing freezes on the main thread after deleting 10,000 objects out of 500,000 on a background thread. Insertions, however, don't cause this issue.
The trigger is the Results observer on the main thread.
Is this a bug in Realm or am I missing something?
Here is an example that produces the mentioned behavior:
AppDelegate
var realm: Realm!
var token: NotificationToken?
func application(_ application: UIApplication, didFinishLaunchingWithOptions launchOptions: [UIApplication.LaunchOptionsKey: Any]?) -> Bool {
realm = try! Realm()
token = realm.objects(Item.self).observe { change in
switch change {
case let .update(_, deletions, insertions, modifications):
print("deletions: \(deletions.count)")
print("insertions: \(insertions.count)")
print("modifications: \(modifications.count)")
default:
break
}
}
// addItemsAsync(count: 600000)
deleteItemsAsync(count: 10000)
return true
}
Adding items
func addItemsAsync(count: Int) {
DispatchQueue.global().async {
autoreleasepool {
let realm = try! Realm()
try! realm.write {
for i in 0..<count {
realm.create(Item.self, value: ["id": i])
}
}
}
}
}
Deleting items
func deleteItemsAsync(count: Int) {
DispatchQueue.global().async {
autoreleasepool {
let realm = try! Realm()
let itemsToDelete = realm.objects(Item.self).filter("id < \(count)")
try! realm.write {
realm.delete(itemsToDelete)
}
}
}
}
Item
class Item: Object {
#objc dynamic var id = 0
}
I also noticed that, unlike insertions, a deletion of this sort doesn't simply notify the observer with 10,000 deletions, but instead I get this here once the results are updated on the main thread:
deletions: 20000
insertions: 10000
modifications: 0
This is obviously due to re-ordering. But I would have expected that Realm updates the results in the background and then simply swaps it on the main thread (especially these kind of expensive operations).
Realm not thread safe. If you want to use in different thread you can use ThreadSafeReference.
Related
My app uses CloudKit and I am trying to implement background fetch.
The method in App Delegate calls a method in my main view controller which checks for changes in the CloudKit database.
However, I realise that I am not calling the completion handler correctly, as the closures for the CloudKit will return asynchronously. I am really unsure how best to call the completion handler in the app delegate method once the operation is complete. Can I pass the completion handler through to the view controller method?
App Delegate
func application(_ application: UIApplication, performFetchWithCompletionHandler completionHandler: #escaping (UIBackgroundFetchResult) -> Void) {
DispatchQueue.global(qos: .userInitiated).async {
// Code to get a reference to main view controller
destinationViewController.getZoneChanges()
completionHandler(.newData)
}
}
Main view controller method to get CloudKit changes
// Fetch zone changes (a method in main table view controller)
func getZoneChanges() {
DispatchQueue.global(qos: .userInitiated).async {
let customZone = CKRecordZone(zoneName: "Drugs")
let zoneID = customZone.zoneID
let zoneIDs = [zoneID]
let changeToken = UserDefaults.standard.serverChangeToken // Custom way of accessing User Defaults using an extension
// Look up the previous change token for each zone
var optionsByRecordZoneID = [CKRecordZone.ID: CKFetchRecordZoneChangesOperation.ZoneOptions]()
// Some other functioning code to process options
// CK Zone Changes Operation
let operation = CKFetchRecordZoneChangesOperation(recordZoneIDs: zoneIDs, optionsByRecordZoneID: optionsByRecordZoneID)
// Closures for records changed, deleted etc.
// Closure details omitted for brevity as fully functional as expected.
// These closures change data model, Spotlight indexing, notifications and trigger UI refresh etc.
operation.recordChangedBlock = { (record) in
// Code...
}
operation.recordWithIDWasDeletedBlock = { (recordId, string) in
// Code...
}
operation.recordZoneChangeTokensUpdatedBlock = { (zoneId, token, data) in
UserDefaults.standard.serverChangeToken = changeToken
UserDefaults.standard.synchronize()
}
operation.recordZoneFetchCompletionBlock = { (zoneId, changeToken, _, _, error) in
if let error = error {
print("Error fetching zone changes: \(error.localizedDescription)")
}
UserDefaults.standard.serverChangeToken = changeToken
UserDefaults.standard.synchronize()
}
operation.fetchRecordZoneChangesCompletionBlock = { (error) in
if let error = error {
print("Error fetching zone changes: \(error.localizedDescription)")
} else {
print("Changes fetched successfully!")
// Save local items
self.saveData() // Uses NSCoding
}
}
CKContainer.default().privateCloudDatabase.add(operation)
}
}
Update your getZoneChanges to have a completion parameter.
func getZoneChanges(completion: #escaping (Bool) -> Void) {
// the rest of your code
operation.fetchRecordZoneChangesCompletionBlock = { (error) in
if let error = error {
print("Error fetching zone changes: \(error.localizedDescription)")
completion(false)
} else {
print("Changes fetched successfully!")
completion(true)
}
}
}
Then you can update the app delegate method to use it:
func application(_ application: UIApplication, performFetchWithCompletionHandler completionHandler: #escaping (UIBackgroundFetchResult) -> Void) {
DispatchQueue.global(qos: .userInitiated).async {
// Code to get a reference to main view controller
destinationViewController.getZoneChanges { (success) in
completionHandler(success ? .newData : .noData)
}
}
}
I have a (custom, linked-list based) queue that I want to deserialize when the app starts and serialize when the app stops, like so (AppDelegate.swift):
func applicationWillResignActive(_ application: UIApplication) {
RequestManager.shared.serializeAndPersistQueue()
}
func applicationDidBecomeActive(_ application: UIApplication) {
RequestManager.shared.deserializeStoredQueue()
}
The issue is during serialization when I exit the app. Here's the code that's running:
public func serializeAndPersistQueue() {
do {
let encoder = JSONEncoder()
let data = try encoder.encode(queue) // Bad access here
if FileManager.default.fileExists(atPath: url.path) {
try FileManager.default.removeItem(at: url)
}
FileManager.default.createFile(atPath: url.path, contents: data, attributes: nil)
}
catch {
print(error)
}
}
As you can see, fairly straightforward. It uses the JSONEncoder to convert my queue to a data object, then writes that data to the file at url.
However, during encoder.encode() I get EXC_BAD_ACCESS every time, without fail.
Additionally, I should note that peak and dequeue operations are conducted on the queue from a background thread. I'm not sure if that makes a difference due to my lack of understanding surrounding GCD. Here's what that method looks like:
private func processRequests() {
DispatchQueue.global(qos: .background).async { [unowned self] in
let group = DispatchGroup()
let semaphore = DispatchSemaphore(value: 0)
while !self.queue.isEmpty {
group.enter()
let request = self.queue.peek()!
self.sendRequest(request: request, completion: { [weak self] in
_ = self?.queue.dequeue()
semaphore.signal()
group.leave()
})
semaphore.wait()
}
group.notify(queue: .global(), execute: { [weak self] in
print("Ending the group")
})
}
}
Lastly, I'll note that:
My queue conforms to the Codable protocol just fine––well, there are no compiler errors, at least. If its implementation beyond that matters, let me know and I'll show it.
The crash occurs a few seconds after I exit the app, while the execution of the processRequests function stops immediately after
So, here's my issue:
The local path for realms on iOS are located in the Documents Directory. I can open them with:
let realm = try! Realm()
Opening a sync realm is different as they are located by URLs
https://realm.io/docs/swift/latest/#realms
I have a UICollectionView with a Results<Object> I can render default data to called in the AppDelegate from a separate file by writing to the Realm on launch
Separate File
class SetUpData {
// MARK: - Seed Realm
static func defaults() {
let realm = try! Realm()
guard realm.isEmpty else { return }
try! realm.write {
realm.add(List.self())
}
}
}
App Delegate
class AppDelegate: UIResponder, UIApplicationDelegate {
var window: UIWindow?
func application(_ application: UIApplication, didFinishLaunchingWithOptions launchOptions: [UIApplicationLaunchOptionsKey: Any]?) -> Bool {
// MARK: - Set Up Realm If Deleted
var config = Realm.Configuration()
config.deleteRealmIfMigrationNeeded = true
Realm.Configuration.defaultConfiguration = config
SetUpData.defaults()
return true
}
From here, client-side (iOS), I am able to successfully log in (rolled my own log in but the values correspond to the Realm Object Server (ROS) admin user) and retrieve the default values from List.cell and begin writing "Lists" to my application.
However, when I configure my realm with a Sync Configuration par opening a synchronized Realm requires a User that’s been authenticated to the Object Server and that’s authorized to open that Realm, I reasonably crash in my cellForItemAtIndexPath return lists.count fatal error: unexpectedly found nil while unwrapping an Optional value because there is no initial data to return.
That makes sense. But what do I do?
Do I need to create a Realm file in the default config and migrate it to the server? I attempted to change my config to a Sync object in the App Delegate with the code below (which is what I am using in ListViewController). No dice.
private func setUpRealm() {
let username = "\(LoginViewController().username.text!)"
let password = "\(LoginViewController().password.text!)"
SyncUser.logIn(with: SyncCredentials.usernamePassword(username: username, password: password, register: true), server: URL(string: "http://000.000.000.000:9080")!) { (user, error) in
guard let user = user else {
fatalError(String(describing: error))
}
DispatchQueue.main.async {
let configuration = Realm.Configuration(syncConfiguration: SyncConfiguration(user: user, realmURL: URL(string: "realm://000.000.000.000:9080/~/realmList")!))
let realm = try! Realm(configuration: configuration)
self.lists = realm.objects(List.self).sorted(byKeyPath: "created", ascending: false)
self.notificationToken = self.lists.addNotificationBlock { [weak self] (changes: RealmCollectionChange) in
guard (self?.collectionView) != nil else { return }
switch changes {
case .initial:
self?.collectionView.reloadData()
break
case .update(_, let deletions, let insertions, let modifications):
self?.collectionView.performBatchUpdates({
self?.collectionView.insertItems(at: insertions.map({ IndexPath(row: $0, section: 0)}))
self?.collectionView.deleteItems(at: deletions.map({ IndexPath(row: $0, section: 0)}))
self?.collectionView.reloadItems(at: modifications.map({ IndexPath(row: $0, section: 0)}))
}, completion: nil)
break
case .error(let error):
print(error.localizedDescription)
break
}
}
}
}
}
Realm does not provide an API to convert a standalone Realm to a synced Realm currently. If my understanding is correct, it is necessary to copy the data from the seed Realm to synced Realm when opening the synced Realm.
I was trying to fetch realm data on the background thread and add a notification block (iOS, Swift).
Basic example:
func initNotificationToken() {
DispatchQueue.global(qos: .background).async {
let realm = try! Realm()
results = self.getRealmResults()
notificationToken = results.addNotificationBlock { [weak self] (changes: RealmCollectionChange) in
switch changes {
case .initial:
self?.initializeDataSource()
break
case .update(_, let deletions, let insertions, let modifications):
self?.updateDataSource(deletions: deletions, insertions: insertions, modifications: modifications)
break
case .error(let error):
fatalError("\(error)")
break
}
}
}
}
func initializeDataSource() {
// process the result set data
DispatchQueue.main.async(execute: { () -> Void in
// update UI
})
}
func updateDataSource(deletions: [Int], insertions: [Int], modifications: [Int]) {
// process the changes in the result set data
DispatchQueue.main.async(execute: { () -> Void in
// update UI
})
}
When doing this I get
'Can only add notification blocks from within runloops'
I have to do some more extensive processing with the returned data and would like to only go back to the main thread when updating the UI after the processing is done.
Another way would probably to re-fetch the data after any update on the background thread and then do the processing, but it feels like avoidable overhead.
Any suggestions on the best practice to solve this?
To add a notification on a background thread you have to manually run a run loop on that thread and add the notification from within a callout from that run loop:
class Stuff {
var token: NotificationToken? = nil
var notificationRunLoop: CFRunLoop? = nil
func initNotificationToken() {
DispatchQueue.global(qos: .background).async {
// Capture a reference to the runloop so that we can stop running it later
notificationRunLoop = CFRunLoopGetCurrent()
CFRunLoopPerformBlock(notificationRunLoop, CFRunLoopMode.defaultMode.rawValue) {
let realm = try! Realm()
results = self.getRealmResults()
// Add the notification from within a block executed by the
// runloop so that Realm can verify that there is actually a
// runloop running on the current thread
token = results.addNotificationBlock { [weak self] (changes: RealmCollectionChange) in
// ...
}
}
// Run the runloop on this thread until we tell it to stop
CFRunLoopRun()
}
}
deinit {
token?.stop()
if let runloop = notificationRunLoop {
CFRunLoopStop(runloop)
}
}
}
GCD does not use a run loop on its worker threads, so anything based on dispatching blocks to the current thread's run loop (such as Realm's notifications) will never get called. To avoid having notifications silently fail to do anything Realm tries to check for this, which unfortunately requires the awakward PerformBlock dance.
In my iOS app users complete transactions which I need to post back to the server. I've created a function to do this:
static let configurationParam = NSURLSessionConfiguration.defaultSessionConfiguration()
static var manager = Alamofire.Manager(configuration: configurationParam)
func postItemToServer(itemToPost:DemoItem) {
let webServiceCallUrl = "..."
var itemApiModel:[String: AnyObject] = [
"ItemId": 123,
"ItemName": itemToPost.Name!,
//...
]
ApiManager.manager.request(.POST, webServiceCallUrl, parameters: itemApiModel, encoding: .JSON)
.validate()
.responseJSON { response in
switch response.result {
case .Success:
print("post success")
case .Failure:
print("SERVER RESPONSE: \(response.response?.statusCode)")
}
}
}
Currently I call this once a transaction is complete:
//...
if(transactionCompleted!) {
let apiManager = ApiManager()
apiManager.postItemToServer(self.item)
self.senderViewController!.performSegueWithIdentifier("TransactionCompletedSegue", sender: self)
}
//...
Where DemoItem is a CoreData object.
This all works as expected. However I need the ability to retry the POST request if it fails. For example if the network connection is down at the point of trying post to the server I need to automatically post the data once it becomes active again - at which point there may be several DemoItem's which need to be synced.
I'm new to Swift. In a similar Xamarin app I had a status column in my SQLite database which I set to 'AwaitingSync'. I then had an async timer that ran every 30 seconds, queried the DB for any items which had status='AwaitingSync' and then tried to post them if they existed. If it succeed it updated the status in the DB. I could implement something along the same lines here - but I was never really happy with that implementation as I had a DB query every 30 seconds even if nothing had changed.
Finally, it needs to be still work if the app is terminated. For example any items which weren't synced before the app is killed should sync once the app is resumed. What's the best way to approach this?
Edit
Based on Tom's answer I've created the following:
class SyncHelper {
let serialQueue = dispatch_queue_create("com.mycompany.syncqueue", DISPATCH_QUEUE_SERIAL)
let managedContext = (UIApplication.sharedApplication().delegate as! AppDelegate).managedObjectContext
func StartSync() {
//Run on serial queue so it can't be called twice at once
dispatch_async(serialQueue, {
//See if there are any items pending to sync
if let itemsToSync = self.GetItemsToSync() {
//Sync all pending items
for itemToSync in itemsToSync {
self.SyncItemToServer(itemToSync)
}
}
})
}
private func GetItemsToSync() -> [DemoItem]? {
var result:[DemoItem]?
do {
let fetchRequest = NSFetchRequest(entityName: "DemoItem")
fetchRequest.predicate = NSPredicate(format: "awaitingSync = true", argumentArray: nil)
result = try managedContext.executeFetchRequest(fetchRequest) as? [DemoItem]
} catch {
//Handle error...
}
return result
}
private func SyncItemToServer(itemToSync:DemoItem) {
let apiManager = ApiManager()
//Try to post to the server
apiManager.postItemToServer(itemToSync:DemoItem, completionHandler: { (error) -> Void in
if let _ = error {
//An error has occurred - nothing need to happen as it will be picked up when the network is restored
print("Sync failed")
} else {
print("Sync success")
itemToSync.awaitingSync = false
do {
try self.managedContext.save()
} catch {
//Handle error...
}
}
})
}
}
I then call this when ever a transaction is completed:
//...
if(transactionCompleted!) {
let syncHelper = SyncHelper()
syncHelper.StartSync()
}
//...
And then finally I've used Reachability.swift to start the sync every time the network connection resumes:
class AppDelegate: UIResponder, UIApplicationDelegate {
var window: UIWindow?
var reachability:Reachability?
func application(application: UIApplication, didFinishLaunchingWithOptions launchOptions: [NSObject: AnyObject]?) -> Bool {
//...
//Setup the sync for when the network connection resumes
do {
reachability = try Reachability.reachabilityForInternetConnection()
NSNotificationCenter.defaultCenter().addObserver(self,
selector: "reachabilityChanged:",
name: ReachabilityChangedNotification,
object: reachability)
try reachability!.startNotifier()
} catch {
print("Unable to create Reachability")
}
return true
}
func reachabilityChanged(note: NSNotification) {
let reachability = note.object as! Reachability
if reachability.isReachable() {
print("Network reachable")
let syncHelper = SyncHelper()
syncHelper.StartSync()
} else {
print("Not reachable")
}
}
}
This all seems to be working. Is this approach ok and have I missed anything which would improve it? The only gap I can see is if the network connectivity is active however the server throws an error for some reason - I guess I could then add a button for the user to retry any pending items.
Firstly, if your concern is whether the network connection is working, you shouldn't be polling at intervals. You should be using iOS's network reachability API to get notified when the network status changes. Apple provides a simple implementation of this and there are numerous alternative implementations online.
Since a sync status value should be a boolean flag, it's not as if a fetch request is a heavy-duty operation, especially if you use reachability. Not only should the fetch request be fast, you can update the flag after the fact in a single step-- use NSBatchUpdateRequest to set the flag to false on every instance you just sent to the server.
If you want to get the sync status out of the persistent store (not a bad idea since it's metadata), you'll need to maintain your own list of unsynced objects. The best way to do this is by tracking the objectID of the managed objects awaiting sync. That would be something like:
Get the objectID of a newly changed managed object
Convert that to an NSURL using NSManagedObjectID's URIRepresentation() method.
Put the NSURL on a list that you save somewhere, so it'll persist.
You can save the list in a file, in user defaults, or in the persistent store's own metadata.
When it's time to sync, you'd do something like:
Get an NSURL from your list
Convert that into an NSManagedObjectID using managedObjectIDForURIRepresentation(url:NSURL) (which is on NSPersistentStoreCoordinator)
Get the managed object for that ID objectWithID: on NSManagedObjectContext.
Sync that object's data.
Then on a successful sync, remove entries from the list.