Overview

Post

Replies

Boosts

Views

Activity

UITab memory leak
I have the following view hierarchy in my app: [UINavigationController] -> [MainViewController] -> [MyTabBarController] -> [DashboardViewController] In my MainViewController I have a button that pushes the MyTabBarController onto the navigation controllers stack. In the tab bar controller I only have one tab in this example showing the DashboardViewController. That all works fine, and when I tap the back button on MyTabBarController, everything works fine and the MainViewController is shown again. The UI works exactly how I want it, but when I load up the 'Debug Memory Graph' view, I can see that my DashboardViewController is still in memory and it seems the UITab has a reference to it. The MyTabBarController is NOT in memory anymore. MyTabBarController is very simple: class MyTabBarController: UITabBarController { override func viewDidLoad() { super.viewDidLoad() self.mode = .tabSidebar var allTabs:[UITab] = [] let mainTab = UITab(title: "Dashboard", image: UIImage(systemName: "chart.pie"), identifier: "dashboard", viewControllerProvider: { _ in return UINavigationController(rootViewController: DashboardViewController()) }) allTabs.append(mainTab) setTabs(allTabs, animated: false) } } And the DashboardViewController is empty: class DashboardViewController: UIViewController { } The only reason I created as a seperate class in this example is so I can easily see if it's visible in the memory debug view. I have uploaded the simple sample app to GitHub: https://github.com/fwaddle/TabbarMemoryLeakCheck Anyone have any suggestions? Here is a screen grab of the memory debug view showing the UITab having a reference to the DashboardViewController even though MyTabBarController has been dealloc'd:
Topic: UI Frameworks SubTopic: UIKit
9
0
395
5d
Please, Apple. I am begging you. Fix the broken Text-To-Speech in macOS
Every new build of macOS 26 further breaks some part of text-to-speech or voice control. I have filed multiple bug reports on this, yet the situation gets worse, not better, with each new build. I am begging you to fix this! I am disabled and rely on these features to work. Accessibility in macOS is more than 50% of my reasons for choosing Macs over Windows. Here's some of what is currently broken: Speak announcements only works if the Samantha voice is selected. If other voices are selected either the announcements don't speak, or they default to the Samantha voice. This became broken with 26.1. Announce the time only works with some voices. I would like to use Australian English Siri 2 but with that voice selected it defaults to Samantha. This became broken with 26.4. With voice control enabled there are two menu bar icons. The blue voice control icon and an orange microphone "an application is accessing the microphone" icon. This orange icon started appearing with 26.3. For four years of macOS releases, the orange warning didn't apply to system services. And note that with voice control enabled on iOS there is no orange icon. It wastes valuable menu bar space and defeats its purpose. With that orange icon always being there I have no indication if a nefarious app starts recording me. This became broken with 26.3. Overlay shows numbers even when it is set to none. This has been broken since at least 14.0. I don't remember if it was broken in prior versions but it has been broken in every version since 14.0. The voice control control center widget is defective. If voice control is not in the menu bar (for example if I've said "Siri turn off voice control") using the control center widget to turn it on brings about the orange icon, but not the blue icon actually used for controlling voice control. If you do have the blue voice control icon and use control center to turn off voice control the blue icon stays but voice control is not enabled. This became broken in 26.0, was fixed in 26.2, broke again in 26.4. When using voice control to edit text (aka dictation mode) saying "go to the end of the line" invariably goes to the beginning of the line. Once in a blue moon it will go to the end, but there is no rhyme or reason and it's rare that it does. Since this bug was added it has worked correctly exactly twice. This became broken in 26.3 (possibly 26.4). I know I am missing some issues. Voice control and text-to-speech have new bugs with each new build of macOS 26. My main Mac is being repaired and once I get it back I'll be installing macOS 15 Sequoia on it because of these issues. These issues stop me from buying a new Mac because any new Mac will only run the broken macOS 26. I file bug reports on each build when I discover another new issue, but these reports seemingly go unread. I would suggest Apple get a focus group of disabled people together and do research into how we use macOS. Find what's broken and what works. And if Apple does this I would be glad to be a part of that group. My place, or yours. Accessibility at one time was something Apple was proud of. It was some Apple showed off. But now, I'm not so sure. It's starting to look like Apple doesn't care. I hope I'm wrong, and that Apple does care, so... Please, Apple. I am begging you. Fix broken Text-To-Speech and Voice Control!
9
2
2.6k
1w
Bug? SwiftData + inheritance + optional many-to-one relationship
I've spent a few months writing an app that uses SwiftData with inheritance. Everything worked well until I tried adding CloudKit support. To do so, I had to make all relationships optional, which exposed what appears to be a bug. Note that this isn't a CloudKit issue -- it happens even when CloudKit is disabled -- but it's due to the requirement for optional relationships. In the code below, I get the following error on the second call to modelContext.save() when the button is clicked: Could not cast value of type 'SwiftData.PersistentIdentifier' (0x1ef510b68) to 'SimplePersistenceIdentifierTest.Computer' (0x1025884e0). I was surprised to find zero hit when Googling "Could not cast value of type 'SwiftData.PersistentIdentifier'". Some things to note: Calling teacher.computers?.append(computer) instead of computer.teacher = teacher results in the same error. It only happens when Teacher inherits Person. It only happens if modelContext.save() is called both times. It works if the first modelContext.save() is commented out. If the second modelContext.save()is commented out, the error occurs the second time the model context is saved (whether explicitly or implicitly). Keep in mind this is a super simple repro written to generate on demand the error I'm seeing in a normal app. In my app, modelContext.save() must be called in some places to update the UI immediately, sometimes resulting in the error seconds later when the model context is saved automatically. Not calling modelContext.save() doesn't appear to be an option. To be sure, I'm new to this ecosystem so I'd be thrilled if I've missed something obvious! Any thoughts are appreciated. import Foundation import SwiftData import SwiftUI struct ContentView: View { @Environment(\.modelContext) var modelContext var body: some View { VStack { Button("Do it") { let teacher = Teacher() let computer = Computer() modelContext.insert(teacher) modelContext.insert(computer) try! modelContext.save() computer.teacher = teacher try! modelContext.save() } } } } @Model class Computer { @Relationship(deleteRule: .nullify) var teacher: Teacher? init() {} } @Model class Person { init() {} } @available(iOS 26.0, macOS 26.0, *) @Model class Teacher: Person { @Relationship(deleteRule: .nullify, inverse: \Computer.teacher) public var computers: [Computer]? = [] override init() { super.init() } }
9
2
486
2w
SwiftData document-based app broken
Hello all Synopsis: document based SwiftData app breaks document handling after first save due to internal error saving the -shm file. Long: i am working on a small document based SwiftData app for macOS. The UI works well as long as the document was not saved. After saving the document and reopening it, I get an error consistently in console: BUG IN CLIENT OF libsqlite3.dylib: database integrity compromised by API violation: vnode unlinked while in use: /Users/vrunkel/Library/Containers/de.ecoobs.CurtailmentAnalyzer/Data/tmp/TemporaryItems/NSIRD_CurtailmentAnalyzer_mrXKMs/NewDocument/StoreContent-shm So somehow the -shm file is still referenced to NewDocument created when the app opens an untitled document and resides in the temporary folder. I have saved the document to my documents folder. After reopening and the above error deletion or addition of items crashes the app with a long backtrace to view updating: Modifications to the layout engine must not be performed from a background thread after it has been accessed from the main thread. I am not creating any threads or do background work. If I do not save the document but work within the new untitled document no problems occur. Even closing the app and reopening the untitled new doc (happens automatically) all is fine. To rule out any influence of my existing view structure I have created the most simple test case - Xcode -> New Project -> macOS document based app configured to use SwiftData. Same behaviour. After saving a new document the addition/deletion of items causes the thread-induced crash and shows the error in console when opening the document. I am using latest versions of Xcode 15.0 and macOS 14.0 Any ideas? thx, volker
9
2
2.5k
5d
Button in ToolbarItem is not completely tapable on iOS 26
I have an icon button in toolbar but only the icon is triggering tap events while outside icon button gives tap feedback but event is not firing. Code: ToolbarItem(placement: .navigationBarTrailing) { Button(action: toggleBookmark) { Image(systemName: isBookmarked ? "bookmark.fill" : "bookmark") .resizable() .aspectRatio(0.8, contentMode: .fit) .frame(width: 20, height: 20) } } Here toggleBookmark function is only called if I click on Image but not if I click outside image but on the circular button that appears on iOS 26. See this screen recording.
Topic: UI Frameworks SubTopic: SwiftUI
8
2
349
2w
good morning My app has been approved and in the app store but my subscriptions are not working
they say developer action needed but I dont understand what they mean, they say I need to attach them to the binary but its already in the appstore and locked I presume, I am tring to create a second version (1.0.1) however I cant find the spot to add the subscriptions because I think they still need developer action kinda feel like I am in a loop and would appreciate some help. I have been using chatgpt but it does not seem to be able to figure it out either, hoping for some good old fashion human help :) thanks again for all that you do, excited to be in the app store and I know this is probably a simple config thing
8
0
297
2w
Is this normal while developing an app on the watch?
I used to develop iPhone apps about a decade ago and am now back at it with an iPhone and companion Apple Watch app. Why does it suck so bad? I a must be doing something wrong, but what I am doing is out of the box. The majority of the time when I go to build on the watch I get: Previous preparation error: A connection to this device could not be established.; Timed out while attempting to establish tunnel using negotiated network parameters. Which I then first try to force quit the app on the watch and if that does not work, reboot the watch (which is pretty ridiculous). I tried Gemini and it gave me the following explanation: This is arguably the single most frustrating error in the entire watchOS development ecosystem. As an expert, I can tell you that you are not doing anything wrong; you are fighting a fragile architecture. Here is the technical breakdown of why this happens and a workflow to mitigate it without constant reboots. The Technical "Why" This error message reveals that the issue lies in the CoreDevice (and formerly usbmuxd) infrastructure, specifically in the secure transport layer. The "Tunnel" Failure: Xcode attempts to establish a secure, encrypted tunnel (DTXConnection) to the Apple Watch to push the debugger and the app binary. Unlike the iPhone, which can sustain a high-bandwidth connection easily, the Watch aggressively manages its radios (Bluetooth vs. WiFi vs. AWDL) to save battery. The "Handshake" Timeout: If the Watch is on Bluetooth (Low Energy) when Xcode initiates the tunnel, the bandwidth is insufficient. The system tries to escalate the connection to WiFi (or Apple Wireless Direct Link). If this negotiation takes even a few milliseconds too long—due to 2.4GHz interference, router mDNS configuration, or the Watch CPU being throttled—the handshake times out. The "Previous Preparation Error": This is the insidious part. Once the handshake fails once, Xcode’s device service caches that failure state for the device UUID. It marks the device as "unreachable" to prevent the IDE from hanging indefinitely on subsequent tries. Why you see it constantly: You aren't actually triggering a new connection attempt; you are hitting the cached "failed" state immediately. Unfortunately, the solutions it gives me are: The "Wired Proxy" Method (Most Reliable) This is the gold standard for watchOS debugging. Do not rely on "Connect via Network" for the Watch directly if you can avoid it. Disable WiFi on your Mac (temporarily) or ensure the Mac and iPhone are on the exact same SSID. Plug your iPhone into the Mac via USB. Ensure the Watch is paired to that iPhone. Result: Xcode will tunnel the instructions through the USB connection to the Phone, and the Phone acts as a high-bandwidth proxy to the Watch. This eliminates the "Mac-to-Watch" WiFi negotiation failure point. Do you hit this regularly? What do you do to make for a smooth development and deployment process? Or is it really this fragile? Thanks for any help! Bryan
8
1
674
3w
Incorrect system color on popover view, and does not update while switching dark mode on iOS 26 beta 3
All system colors are displayed incorrectly on the popover view. Those are the same views present as a popover in light and dark mode. And those are the same views present as modal. And there is also a problem that when the popover is presented, switching to dark/light mode will not change the appearance. That affected all system apps. The following screenshot is already in dark mode. All those problem are occured on iOS 26 beta 3.
8
0
1k
2w
MFMailComposeViewController in visionOS does not have a cancel button
When i use the MFMailComposeViewController in visionOS, there is no cancel button for the controller. The button at the bottom closes the app. Is anyone else experiencing this? if([MFMailComposeViewController canSendMail]) { MFMailComposeViewController* controller = [[MFMailComposeViewController alloc] init]; controller.mailComposeDelegate = (id <MFMailComposeViewControllerDelegate>)view; [controller setToRecipients:toAddresses]; [controller setSubject:subject]; [controller setMessageBody:body isHTML:isHtml]; [view presentViewController:controller animated:YES completion:nil]; }
8
0
757
1w
XCTest Bundle cannot access local network.
We’re having an iPad issue accessing the local network with iPadOS 26.3. We have an automation system that tests our app on an iPad using accessibility tags. the XCTest test code sends messages from the iPad via TCP/IP to setup external test equipment. The messages abruptly stopped transmitting across the iPad blood-brain barrier with iPadOS 26.3 (26.2.1 and earlier works fine). The technique that worked involved installing a helper app with the same bundleID as our app, allowing the helper app to access the network, and when our app runs it has network access through the helper. It’s clever and kludgey. Forums that we referenced in the past: https://aninterestingwebsite.com/forums/thread/663858 TN3179: Understanding local network privacy | Apple Developer Documentation I suspect that something was changed in 26.3 that closed our window. I need two things: ID what is different in 26.3 and fix the automation system. If there’s a new way for XCUITest code to access the local network I’m happy to try it out.
8
0
167
3w
Playgrounds app with the latest SDK (OS 26)
I am preparing for Swift Student Challenge, but it seems like the iPad Playground app still not support the latest SDK. I can't use frameworks like FoundationModel, etc., directly in playground app My playground for this year would require iPad environment since it uses PencilKit, ARKit, etc., and I also want to use the latest tech + the liquid glass UI Right now, I develop the project as a normal Xcode project, and I am wondering how do I post it? As Xcode playground, it must run on macOS As Swift Playground project, the iPad version of Playground does not support latest APIs and it can't compile
8
5
1.4k
2w
Push Notifications
The following issue has occurred: Push notifications are not being received on certain devices. What could be the possible causes? Push notifications are being sent from our own server, and we are receiving normal responses from APNs. Users have confirmed that notifications are enabled on their devices, and they report no network issues. This problem is occurring for multiple users.
8
0
359
2w
Provisioning profile missing entitlement: com.apple.developer.icloud
Hi, I do have a strange behavior in my development environment on a Mac mini (M4) running 26.2 and Xcode 26.3. Everything was working as expected. My project had a stable state and I wanted to enable iCloud support. As result I could not run the app any more because code signing failed with the message that my profile does not include the above entitlement. On my notebook (M2) with XCode 26.3 everything is working. Im am using GIT and both computers have identical code. The code compiling and running on my notebook will not run any more on my Max mini. Any help to find what might have broken the code signing and how it could be fixed? Thanks in advance.
8
0
300
2w
iOS 18 new RecognizedTextRequest DEADLOCKS if more than 2 are run in parallel
Following WWDC24 video "Discover Swift enhancements in the Vision framework" recommendations (cfr video at 10'41"), I used the following code to perform multiple new iOS 18 `RecognizedTextRequest' in parallel. Problem: if more than 2 request are run in parallel, the request will hang, leaving the app in a state where no more requests can be started. -> deadlock I tried other ways to run the requests, but no matter the method employed, or what device I use: no more than 2 requests can ever be run in parallel. func triggerDeadlock() {} try await withThrowingTaskGroup(of: Void.self) { group in // See: WWDC 2024 Discover Siwft enhancements in the Vision framework at 10:41 // ############## THIS IS KEY let maxOCRTasks = 5 // On a real-device, if more than 2 RecognizeTextRequest are launched in parallel using tasks, the request hangs // ############## THIS IS KEY for idx in 0..<maxOCRTasks { let url = ... // URL to some image group.addTask { // Perform OCR let _ = await performOCRRequest(on: url: url) } } var nextIndex = maxOCRTasks for try await _ in group { // Wait for the result of the next child task that finished if nextIndex < pageCount { group.addTask { let url = ... // URL to some image // Perform OCR let _ = await performOCRRequest(on: url: url) } nextIndex += 1 } } } } // MARK: - ASYNC/AWAIT version with iOS 18 @available(iOS 18, *) func performOCRRequest(on url: URL) async throws -> [RecognizedText] { // Create request var request = RecognizeTextRequest() // Single request: no need for ImageRequestHandler // Configure request request.recognitionLevel = .accurate request.automaticallyDetectsLanguage = true request.usesLanguageCorrection = true request.minimumTextHeightFraction = 0.016 // Perform request let textObservations: [RecognizedTextObservation] = try await request.perform(on: url) // Convert [RecognizedTextObservation] to [RecognizedText] return textObservations.compactMap { observation in observation.topCandidates(1).first } } I also found this Swift forums post mentioning something very similar. I also opened a feedback: FB17240843
8
0
512
1w
App group broken on Sequoia
We've got an in-house Swift macOS desktop app with a FileProvider extension, which has been working fine on various machines up through Sonoma (and still does). We've just installed it on a Sequoia machine, and on it the FileProvider extension has lost the ability to access the shared app group. It can neither log to the Group Containers folder under ~/Library, nor access the pipe to the main app. The group name is formatted as group.XXXXXXXXXX.com.orgname.appname in both targets. I'm not sure why it combines the iOS and macOS conventions, with both the group prefix and the teamIdentifier one -- it was first built some time before the point in 2025 when macOS supported iOS-style groups -- but again, it's been working. For the record, The provisioning profile for EMPFileProvider has the App Groups capability enabled, and the App Groups capability is present in both build targets in Xcode. The existing group identifier is registered on the website; I've also manually registered the team-ID-less group name, so I can migrate. The question is, is this actually the right approach? Will such a change break the app on pre-Sequoia machines? And if I proceed, what do I need to do to complete the migration? The app was built back in Xcode 12.5; will I need to update the entire build environment to take advantage of Xcode 16.3's explicit support for iOS-style group names, or can I get away with it since I've manually registered the new group?
8
0
315
3w
iOS 26.1 PHPickerConfiguration.preselectedAssetIdentifiers doesn't select previous pictures in the PHPickerViewController
Hi, I faced with the issue on iOS 26.1 with PHPickerViewController. After first selection I save assetIdentifier of PHPickerResult for images. next time I open the picker I expect to have the images selected based on assetIdentifier Code: var config = PHPickerConfiguration(photoLibrary: .shared()) config.selectionLimit = 10 config.filter = .images config.preselectedAssetIdentifiers = images.compactMap(\.assetID) let picker = PHPickerViewController(configuration: config) picker.delegate = self present(picker, animated: true) But on iOS 26.1 they aren't selected. On lower iOS version all works fine. Does anybody faced with similar issue?
Topic: UI Frameworks SubTopic: UIKit
8
7
695
1w
CoreML regression between macOS 26.0.1 and macOS 26.1 Beta causing scrambled tensor outputs
We’ve encountered what appears to be a CoreML regression between macOS 26.0.1 and macOS 26.1 Beta. In macOS 26.0.1, CoreML models run and produce correct results. However, in macOS 26.1 Beta, the same models produce scrambled or corrupted outputs, suggesting that tensor memory is being read or written incorrectly. The behavior is consistent with a low-level stride or pointer arithmetic issue — for example, using 16-bit strides on 32-bit data or other mismatches in tensor layout handling. Reproduction Install ON1 Photo RAW 2026 or ON1 Resize 2026 on macOS 26.0.1. Use the newest Highest Quality resize model, which is Stable Diffusion–based and runs through CoreML. Observe correct, high-quality results. Upgrade to macOS 26.1 Beta and run the same operation again. The output becomes visually scrambled or corrupted. We are also seeing similar issues with another Stable Diffusion UNet model that previously worked correctly on macOS 26.0.1. This suggests the regression may affect multiple diffusion-style architectures, likely due to a change in CoreML’s tensor stride, layout computation, or memory alignment between these versions. Notes The affected models are exported using standard CoreML conversion pipelines. No custom operators or third-party CoreML runtime layers are used. The issue reproduces consistently across multiple machines. It would be helpful to know if there were changes to CoreML’s tensor layout, precision handling, or MLCompute backend between macOS 26.0.1 and 26.1 Beta, or if this is a known regression in the current beta.
8
4
2.3k
1w
Why do random errOSAInternalTableOverflow errors return when running AppleScripts via ScriptingBridge?
We have an app that controls InDesign Desktop and InDesignServer via hundreds of AppleScripts. Some macOS security updates a while back dictated that we start communicating with other apps via ScriptingBridge. We couldn't afford to convert the hundreds of AppleScripts into direct ScriptingBridge nomenclature, so we opted to keep them as is and instead tell the external apps to: [app doScript:<the script text> language:InDesignScLgApplescriptLanguage withArguments:nil undoMode:InDesignESUMScriptRequest undoName:@"blah"] There are a handful of scripts that we did convert to direct ScriptingBridge. There are times (and under the right circumstances, it's repeatable) when a certain script will have run perfectly dozens of times, and then it will throw errOSAInternalTableOverflow. We create a new SBApplication for every job (which could be a single instance of Desktop or the multiple instances of Server). Why is this error happening seemingly randomly? Is there anything we can do to work around or prevent this?
8
0
308
4d
UITab memory leak
I have the following view hierarchy in my app: [UINavigationController] -> [MainViewController] -> [MyTabBarController] -> [DashboardViewController] In my MainViewController I have a button that pushes the MyTabBarController onto the navigation controllers stack. In the tab bar controller I only have one tab in this example showing the DashboardViewController. That all works fine, and when I tap the back button on MyTabBarController, everything works fine and the MainViewController is shown again. The UI works exactly how I want it, but when I load up the 'Debug Memory Graph' view, I can see that my DashboardViewController is still in memory and it seems the UITab has a reference to it. The MyTabBarController is NOT in memory anymore. MyTabBarController is very simple: class MyTabBarController: UITabBarController { override func viewDidLoad() { super.viewDidLoad() self.mode = .tabSidebar var allTabs:[UITab] = [] let mainTab = UITab(title: "Dashboard", image: UIImage(systemName: "chart.pie"), identifier: "dashboard", viewControllerProvider: { _ in return UINavigationController(rootViewController: DashboardViewController()) }) allTabs.append(mainTab) setTabs(allTabs, animated: false) } } And the DashboardViewController is empty: class DashboardViewController: UIViewController { } The only reason I created as a seperate class in this example is so I can easily see if it's visible in the memory debug view. I have uploaded the simple sample app to GitHub: https://github.com/fwaddle/TabbarMemoryLeakCheck Anyone have any suggestions? Here is a screen grab of the memory debug view showing the UITab having a reference to the DashboardViewController even though MyTabBarController has been dealloc'd:
Topic: UI Frameworks SubTopic: UIKit
Replies
9
Boosts
0
Views
395
Activity
5d
Please, Apple. I am begging you. Fix the broken Text-To-Speech in macOS
Every new build of macOS 26 further breaks some part of text-to-speech or voice control. I have filed multiple bug reports on this, yet the situation gets worse, not better, with each new build. I am begging you to fix this! I am disabled and rely on these features to work. Accessibility in macOS is more than 50% of my reasons for choosing Macs over Windows. Here's some of what is currently broken: Speak announcements only works if the Samantha voice is selected. If other voices are selected either the announcements don't speak, or they default to the Samantha voice. This became broken with 26.1. Announce the time only works with some voices. I would like to use Australian English Siri 2 but with that voice selected it defaults to Samantha. This became broken with 26.4. With voice control enabled there are two menu bar icons. The blue voice control icon and an orange microphone "an application is accessing the microphone" icon. This orange icon started appearing with 26.3. For four years of macOS releases, the orange warning didn't apply to system services. And note that with voice control enabled on iOS there is no orange icon. It wastes valuable menu bar space and defeats its purpose. With that orange icon always being there I have no indication if a nefarious app starts recording me. This became broken with 26.3. Overlay shows numbers even when it is set to none. This has been broken since at least 14.0. I don't remember if it was broken in prior versions but it has been broken in every version since 14.0. The voice control control center widget is defective. If voice control is not in the menu bar (for example if I've said "Siri turn off voice control") using the control center widget to turn it on brings about the orange icon, but not the blue icon actually used for controlling voice control. If you do have the blue voice control icon and use control center to turn off voice control the blue icon stays but voice control is not enabled. This became broken in 26.0, was fixed in 26.2, broke again in 26.4. When using voice control to edit text (aka dictation mode) saying "go to the end of the line" invariably goes to the beginning of the line. Once in a blue moon it will go to the end, but there is no rhyme or reason and it's rare that it does. Since this bug was added it has worked correctly exactly twice. This became broken in 26.3 (possibly 26.4). I know I am missing some issues. Voice control and text-to-speech have new bugs with each new build of macOS 26. My main Mac is being repaired and once I get it back I'll be installing macOS 15 Sequoia on it because of these issues. These issues stop me from buying a new Mac because any new Mac will only run the broken macOS 26. I file bug reports on each build when I discover another new issue, but these reports seemingly go unread. I would suggest Apple get a focus group of disabled people together and do research into how we use macOS. Find what's broken and what works. And if Apple does this I would be glad to be a part of that group. My place, or yours. Accessibility at one time was something Apple was proud of. It was some Apple showed off. But now, I'm not so sure. It's starting to look like Apple doesn't care. I hope I'm wrong, and that Apple does care, so... Please, Apple. I am begging you. Fix broken Text-To-Speech and Voice Control!
Replies
9
Boosts
2
Views
2.6k
Activity
1w
Bug? SwiftData + inheritance + optional many-to-one relationship
I've spent a few months writing an app that uses SwiftData with inheritance. Everything worked well until I tried adding CloudKit support. To do so, I had to make all relationships optional, which exposed what appears to be a bug. Note that this isn't a CloudKit issue -- it happens even when CloudKit is disabled -- but it's due to the requirement for optional relationships. In the code below, I get the following error on the second call to modelContext.save() when the button is clicked: Could not cast value of type 'SwiftData.PersistentIdentifier' (0x1ef510b68) to 'SimplePersistenceIdentifierTest.Computer' (0x1025884e0). I was surprised to find zero hit when Googling "Could not cast value of type 'SwiftData.PersistentIdentifier'". Some things to note: Calling teacher.computers?.append(computer) instead of computer.teacher = teacher results in the same error. It only happens when Teacher inherits Person. It only happens if modelContext.save() is called both times. It works if the first modelContext.save() is commented out. If the second modelContext.save()is commented out, the error occurs the second time the model context is saved (whether explicitly or implicitly). Keep in mind this is a super simple repro written to generate on demand the error I'm seeing in a normal app. In my app, modelContext.save() must be called in some places to update the UI immediately, sometimes resulting in the error seconds later when the model context is saved automatically. Not calling modelContext.save() doesn't appear to be an option. To be sure, I'm new to this ecosystem so I'd be thrilled if I've missed something obvious! Any thoughts are appreciated. import Foundation import SwiftData import SwiftUI struct ContentView: View { @Environment(\.modelContext) var modelContext var body: some View { VStack { Button("Do it") { let teacher = Teacher() let computer = Computer() modelContext.insert(teacher) modelContext.insert(computer) try! modelContext.save() computer.teacher = teacher try! modelContext.save() } } } } @Model class Computer { @Relationship(deleteRule: .nullify) var teacher: Teacher? init() {} } @Model class Person { init() {} } @available(iOS 26.0, macOS 26.0, *) @Model class Teacher: Person { @Relationship(deleteRule: .nullify, inverse: \Computer.teacher) public var computers: [Computer]? = [] override init() { super.init() } }
Replies
9
Boosts
2
Views
486
Activity
2w
SwiftData document-based app broken
Hello all Synopsis: document based SwiftData app breaks document handling after first save due to internal error saving the -shm file. Long: i am working on a small document based SwiftData app for macOS. The UI works well as long as the document was not saved. After saving the document and reopening it, I get an error consistently in console: BUG IN CLIENT OF libsqlite3.dylib: database integrity compromised by API violation: vnode unlinked while in use: /Users/vrunkel/Library/Containers/de.ecoobs.CurtailmentAnalyzer/Data/tmp/TemporaryItems/NSIRD_CurtailmentAnalyzer_mrXKMs/NewDocument/StoreContent-shm So somehow the -shm file is still referenced to NewDocument created when the app opens an untitled document and resides in the temporary folder. I have saved the document to my documents folder. After reopening and the above error deletion or addition of items crashes the app with a long backtrace to view updating: Modifications to the layout engine must not be performed from a background thread after it has been accessed from the main thread. I am not creating any threads or do background work. If I do not save the document but work within the new untitled document no problems occur. Even closing the app and reopening the untitled new doc (happens automatically) all is fine. To rule out any influence of my existing view structure I have created the most simple test case - Xcode -> New Project -> macOS document based app configured to use SwiftData. Same behaviour. After saving a new document the addition/deletion of items causes the thread-induced crash and shows the error in console when opening the document. I am using latest versions of Xcode 15.0 and macOS 14.0 Any ideas? thx, volker
Replies
9
Boosts
2
Views
2.5k
Activity
5d
Slow launch of app on iOS Simulator 26.4
Each time I launch a Debug version of the app on iOS Simulator, I see the Launch Screen presented and then about a 30-second delay before the app is responsive and debug output appears in the console panel. Filed FB22345091
Replies
8
Boosts
3
Views
402
Activity
6d
Button in ToolbarItem is not completely tapable on iOS 26
I have an icon button in toolbar but only the icon is triggering tap events while outside icon button gives tap feedback but event is not firing. Code: ToolbarItem(placement: .navigationBarTrailing) { Button(action: toggleBookmark) { Image(systemName: isBookmarked ? "bookmark.fill" : "bookmark") .resizable() .aspectRatio(0.8, contentMode: .fit) .frame(width: 20, height: 20) } } Here toggleBookmark function is only called if I click on Image but not if I click outside image but on the circular button that appears on iOS 26. See this screen recording.
Topic: UI Frameworks SubTopic: SwiftUI
Replies
8
Boosts
2
Views
349
Activity
2w
good morning My app has been approved and in the app store but my subscriptions are not working
they say developer action needed but I dont understand what they mean, they say I need to attach them to the binary but its already in the appstore and locked I presume, I am tring to create a second version (1.0.1) however I cant find the spot to add the subscriptions because I think they still need developer action kinda feel like I am in a loop and would appreciate some help. I have been using chatgpt but it does not seem to be able to figure it out either, hoping for some good old fashion human help :) thanks again for all that you do, excited to be in the app store and I know this is probably a simple config thing
Replies
8
Boosts
0
Views
297
Activity
2w
Is this normal while developing an app on the watch?
I used to develop iPhone apps about a decade ago and am now back at it with an iPhone and companion Apple Watch app. Why does it suck so bad? I a must be doing something wrong, but what I am doing is out of the box. The majority of the time when I go to build on the watch I get: Previous preparation error: A connection to this device could not be established.; Timed out while attempting to establish tunnel using negotiated network parameters. Which I then first try to force quit the app on the watch and if that does not work, reboot the watch (which is pretty ridiculous). I tried Gemini and it gave me the following explanation: This is arguably the single most frustrating error in the entire watchOS development ecosystem. As an expert, I can tell you that you are not doing anything wrong; you are fighting a fragile architecture. Here is the technical breakdown of why this happens and a workflow to mitigate it without constant reboots. The Technical "Why" This error message reveals that the issue lies in the CoreDevice (and formerly usbmuxd) infrastructure, specifically in the secure transport layer. The "Tunnel" Failure: Xcode attempts to establish a secure, encrypted tunnel (DTXConnection) to the Apple Watch to push the debugger and the app binary. Unlike the iPhone, which can sustain a high-bandwidth connection easily, the Watch aggressively manages its radios (Bluetooth vs. WiFi vs. AWDL) to save battery. The "Handshake" Timeout: If the Watch is on Bluetooth (Low Energy) when Xcode initiates the tunnel, the bandwidth is insufficient. The system tries to escalate the connection to WiFi (or Apple Wireless Direct Link). If this negotiation takes even a few milliseconds too long—due to 2.4GHz interference, router mDNS configuration, or the Watch CPU being throttled—the handshake times out. The "Previous Preparation Error": This is the insidious part. Once the handshake fails once, Xcode’s device service caches that failure state for the device UUID. It marks the device as "unreachable" to prevent the IDE from hanging indefinitely on subsequent tries. Why you see it constantly: You aren't actually triggering a new connection attempt; you are hitting the cached "failed" state immediately. Unfortunately, the solutions it gives me are: The "Wired Proxy" Method (Most Reliable) This is the gold standard for watchOS debugging. Do not rely on "Connect via Network" for the Watch directly if you can avoid it. Disable WiFi on your Mac (temporarily) or ensure the Mac and iPhone are on the exact same SSID. Plug your iPhone into the Mac via USB. Ensure the Watch is paired to that iPhone. Result: Xcode will tunnel the instructions through the USB connection to the Phone, and the Phone acts as a high-bandwidth proxy to the Watch. This eliminates the "Mac-to-Watch" WiFi negotiation failure point. Do you hit this regularly? What do you do to make for a smooth development and deployment process? Or is it really this fragile? Thanks for any help! Bryan
Replies
8
Boosts
1
Views
674
Activity
3w
Incorrect system color on popover view, and does not update while switching dark mode on iOS 26 beta 3
All system colors are displayed incorrectly on the popover view. Those are the same views present as a popover in light and dark mode. And those are the same views present as modal. And there is also a problem that when the popover is presented, switching to dark/light mode will not change the appearance. That affected all system apps. The following screenshot is already in dark mode. All those problem are occured on iOS 26 beta 3.
Replies
8
Boosts
0
Views
1k
Activity
2w
Waiting for review?
My first App, a very simple one, is currently ‘Waiting for Review’. Is there anything else I need to do please, or do I just need to wait? Thanks
Replies
8
Boosts
0
Views
313
Activity
2w
MFMailComposeViewController in visionOS does not have a cancel button
When i use the MFMailComposeViewController in visionOS, there is no cancel button for the controller. The button at the bottom closes the app. Is anyone else experiencing this? if([MFMailComposeViewController canSendMail]) { MFMailComposeViewController* controller = [[MFMailComposeViewController alloc] init]; controller.mailComposeDelegate = (id <MFMailComposeViewControllerDelegate>)view; [controller setToRecipients:toAddresses]; [controller setSubject:subject]; [controller setMessageBody:body isHTML:isHtml]; [view presentViewController:controller animated:YES completion:nil]; }
Replies
8
Boosts
0
Views
757
Activity
1w
XCTest Bundle cannot access local network.
We’re having an iPad issue accessing the local network with iPadOS 26.3. We have an automation system that tests our app on an iPad using accessibility tags. the XCTest test code sends messages from the iPad via TCP/IP to setup external test equipment. The messages abruptly stopped transmitting across the iPad blood-brain barrier with iPadOS 26.3 (26.2.1 and earlier works fine). The technique that worked involved installing a helper app with the same bundleID as our app, allowing the helper app to access the network, and when our app runs it has network access through the helper. It’s clever and kludgey. Forums that we referenced in the past: https://aninterestingwebsite.com/forums/thread/663858 TN3179: Understanding local network privacy | Apple Developer Documentation I suspect that something was changed in 26.3 that closed our window. I need two things: ID what is different in 26.3 and fix the automation system. If there’s a new way for XCUITest code to access the local network I’m happy to try it out.
Replies
8
Boosts
0
Views
167
Activity
3w
Playgrounds app with the latest SDK (OS 26)
I am preparing for Swift Student Challenge, but it seems like the iPad Playground app still not support the latest SDK. I can't use frameworks like FoundationModel, etc., directly in playground app My playground for this year would require iPad environment since it uses PencilKit, ARKit, etc., and I also want to use the latest tech + the liquid glass UI Right now, I develop the project as a normal Xcode project, and I am wondering how do I post it? As Xcode playground, it must run on macOS As Swift Playground project, the iPad version of Playground does not support latest APIs and it can't compile
Replies
8
Boosts
5
Views
1.4k
Activity
2w
Push Notifications
The following issue has occurred: Push notifications are not being received on certain devices. What could be the possible causes? Push notifications are being sent from our own server, and we are receiving normal responses from APNs. Users have confirmed that notifications are enabled on their devices, and they report no network issues. This problem is occurring for multiple users.
Replies
8
Boosts
0
Views
359
Activity
2w
Provisioning profile missing entitlement: com.apple.developer.icloud
Hi, I do have a strange behavior in my development environment on a Mac mini (M4) running 26.2 and Xcode 26.3. Everything was working as expected. My project had a stable state and I wanted to enable iCloud support. As result I could not run the app any more because code signing failed with the message that my profile does not include the above entitlement. On my notebook (M2) with XCode 26.3 everything is working. Im am using GIT and both computers have identical code. The code compiling and running on my notebook will not run any more on my Max mini. Any help to find what might have broken the code signing and how it could be fixed? Thanks in advance.
Replies
8
Boosts
0
Views
300
Activity
2w
iOS 18 new RecognizedTextRequest DEADLOCKS if more than 2 are run in parallel
Following WWDC24 video "Discover Swift enhancements in the Vision framework" recommendations (cfr video at 10'41"), I used the following code to perform multiple new iOS 18 `RecognizedTextRequest' in parallel. Problem: if more than 2 request are run in parallel, the request will hang, leaving the app in a state where no more requests can be started. -> deadlock I tried other ways to run the requests, but no matter the method employed, or what device I use: no more than 2 requests can ever be run in parallel. func triggerDeadlock() {} try await withThrowingTaskGroup(of: Void.self) { group in // See: WWDC 2024 Discover Siwft enhancements in the Vision framework at 10:41 // ############## THIS IS KEY let maxOCRTasks = 5 // On a real-device, if more than 2 RecognizeTextRequest are launched in parallel using tasks, the request hangs // ############## THIS IS KEY for idx in 0..<maxOCRTasks { let url = ... // URL to some image group.addTask { // Perform OCR let _ = await performOCRRequest(on: url: url) } } var nextIndex = maxOCRTasks for try await _ in group { // Wait for the result of the next child task that finished if nextIndex < pageCount { group.addTask { let url = ... // URL to some image // Perform OCR let _ = await performOCRRequest(on: url: url) } nextIndex += 1 } } } } // MARK: - ASYNC/AWAIT version with iOS 18 @available(iOS 18, *) func performOCRRequest(on url: URL) async throws -> [RecognizedText] { // Create request var request = RecognizeTextRequest() // Single request: no need for ImageRequestHandler // Configure request request.recognitionLevel = .accurate request.automaticallyDetectsLanguage = true request.usesLanguageCorrection = true request.minimumTextHeightFraction = 0.016 // Perform request let textObservations: [RecognizedTextObservation] = try await request.perform(on: url) // Convert [RecognizedTextObservation] to [RecognizedText] return textObservations.compactMap { observation in observation.topCandidates(1).first } } I also found this Swift forums post mentioning something very similar. I also opened a feedback: FB17240843
Replies
8
Boosts
0
Views
512
Activity
1w
App group broken on Sequoia
We've got an in-house Swift macOS desktop app with a FileProvider extension, which has been working fine on various machines up through Sonoma (and still does). We've just installed it on a Sequoia machine, and on it the FileProvider extension has lost the ability to access the shared app group. It can neither log to the Group Containers folder under ~/Library, nor access the pipe to the main app. The group name is formatted as group.XXXXXXXXXX.com.orgname.appname in both targets. I'm not sure why it combines the iOS and macOS conventions, with both the group prefix and the teamIdentifier one -- it was first built some time before the point in 2025 when macOS supported iOS-style groups -- but again, it's been working. For the record, The provisioning profile for EMPFileProvider has the App Groups capability enabled, and the App Groups capability is present in both build targets in Xcode. The existing group identifier is registered on the website; I've also manually registered the team-ID-less group name, so I can migrate. The question is, is this actually the right approach? Will such a change break the app on pre-Sequoia machines? And if I proceed, what do I need to do to complete the migration? The app was built back in Xcode 12.5; will I need to update the entire build environment to take advantage of Xcode 16.3's explicit support for iOS-style group names, or can I get away with it since I've manually registered the new group?
Replies
8
Boosts
0
Views
315
Activity
3w
iOS 26.1 PHPickerConfiguration.preselectedAssetIdentifiers doesn't select previous pictures in the PHPickerViewController
Hi, I faced with the issue on iOS 26.1 with PHPickerViewController. After first selection I save assetIdentifier of PHPickerResult for images. next time I open the picker I expect to have the images selected based on assetIdentifier Code: var config = PHPickerConfiguration(photoLibrary: .shared()) config.selectionLimit = 10 config.filter = .images config.preselectedAssetIdentifiers = images.compactMap(\.assetID) let picker = PHPickerViewController(configuration: config) picker.delegate = self present(picker, animated: true) But on iOS 26.1 they aren't selected. On lower iOS version all works fine. Does anybody faced with similar issue?
Topic: UI Frameworks SubTopic: UIKit
Replies
8
Boosts
7
Views
695
Activity
1w
CoreML regression between macOS 26.0.1 and macOS 26.1 Beta causing scrambled tensor outputs
We’ve encountered what appears to be a CoreML regression between macOS 26.0.1 and macOS 26.1 Beta. In macOS 26.0.1, CoreML models run and produce correct results. However, in macOS 26.1 Beta, the same models produce scrambled or corrupted outputs, suggesting that tensor memory is being read or written incorrectly. The behavior is consistent with a low-level stride or pointer arithmetic issue — for example, using 16-bit strides on 32-bit data or other mismatches in tensor layout handling. Reproduction Install ON1 Photo RAW 2026 or ON1 Resize 2026 on macOS 26.0.1. Use the newest Highest Quality resize model, which is Stable Diffusion–based and runs through CoreML. Observe correct, high-quality results. Upgrade to macOS 26.1 Beta and run the same operation again. The output becomes visually scrambled or corrupted. We are also seeing similar issues with another Stable Diffusion UNet model that previously worked correctly on macOS 26.0.1. This suggests the regression may affect multiple diffusion-style architectures, likely due to a change in CoreML’s tensor stride, layout computation, or memory alignment between these versions. Notes The affected models are exported using standard CoreML conversion pipelines. No custom operators or third-party CoreML runtime layers are used. The issue reproduces consistently across multiple machines. It would be helpful to know if there were changes to CoreML’s tensor layout, precision handling, or MLCompute backend between macOS 26.0.1 and 26.1 Beta, or if this is a known regression in the current beta.
Replies
8
Boosts
4
Views
2.3k
Activity
1w
Why do random errOSAInternalTableOverflow errors return when running AppleScripts via ScriptingBridge?
We have an app that controls InDesign Desktop and InDesignServer via hundreds of AppleScripts. Some macOS security updates a while back dictated that we start communicating with other apps via ScriptingBridge. We couldn't afford to convert the hundreds of AppleScripts into direct ScriptingBridge nomenclature, so we opted to keep them as is and instead tell the external apps to: [app doScript:<the script text> language:InDesignScLgApplescriptLanguage withArguments:nil undoMode:InDesignESUMScriptRequest undoName:@"blah"] There are a handful of scripts that we did convert to direct ScriptingBridge. There are times (and under the right circumstances, it's repeatable) when a certain script will have run perfectly dozens of times, and then it will throw errOSAInternalTableOverflow. We create a new SBApplication for every job (which could be a single instance of Desktop or the multiple instances of Server). Why is this error happening seemingly randomly? Is there anything we can do to work around or prevent this?
Replies
8
Boosts
0
Views
308
Activity
4d