I noticed that trying to access safeAreaInsets from the active window causes an infinite run loop.
This issue appeared after updating to Beta 3.
Here’s an example of the code:
extension UIDevice {
var safeAreaInsets: UIEdgeInsets {
guard let windowScene = UIApplication.shared.connectedScenes.first as? UIWindowScene,
let window = windowScene.windows.first(where: { $0.isKeyWindow }) else {
return .zero
}
return window.safeAreaInsets
}
}
The return doesn’t happen because it ends up in some kind of recursion.
Explore the various UI frameworks available for building app interfaces. Discuss the use cases for different frameworks, share best practices, and get help with specific framework-related questions.
Selecting any option will automatically load the page
Post
Replies
Boosts
Views
Activity
Why is the UIKeyboard implementation still holding a reference to this UITextField, thus keeping it from being deallocated?
The memory debugger shows:
UIKeyboardImpl -> UIKBAutofillController -> NSMutableDictionary -> NSMutable...(Storage) -> UITextField
Any idea what's going on there?
In iOS26, when using a standalone UITabBar without UITabBarController, the liquid glass blur effect is not applied when scrollable content moves behind the tab bar. However, the blur effect appears correctly when using UITabBarController.
Sample Screenshots:
When using UITababr
When using UITababrController
Sample Code:
class SimpleTabBarController: UIViewController, UITabBarDelegate {
let tabBar = UITabBar()
let redItem = UITabBarItem(title: "Red", image: .add, tag: 0)
let blueItem = UITabBarItem(title: "Blue", image: .checkmark, tag: 1)
override func viewDidLoad() {
super.viewDidLoad()
view.backgroundColor = .white
tabBar.items = [redItem, blueItem]
tabBar.selectedItem = redItem
tabBar.delegate = self
tabBar.translatesAutoresizingMaskIntoConstraints = false
let tableContainerView = TableContainerView()
view.addSubview(tableContainerView)
tableContainerView.translatesAutoresizingMaskIntoConstraints = false
NSLayoutConstraint.activate([
tableContainerView.topAnchor.constraint(equalTo: view.topAnchor),
tableContainerView.leadingAnchor.constraint(equalTo: view.leadingAnchor),
tableContainerView.trailingAnchor.constraint(equalTo: view.trailingAnchor),
tableContainerView.bottomAnchor.constraint(equalTo: view.bottomAnchor)
])
view.addSubview(tabBar)
NSLayoutConstraint.activate([
tabBar.leadingAnchor.constraint(equalTo: view.leadingAnchor),
tabBar.trailingAnchor.constraint(equalTo: view.trailingAnchor),
tabBar.bottomAnchor.constraint(equalTo: view.bottomAnchor)
])
}
There is a serious usability issue with PHPickerViewController in a UIKit app running on macOS 26 via Mac Catalyst when the Mac Catalyst interface is set to “Scaled to Match iPad”. Mouse click and other pointer interactions do not take place in the correct position. This means you have to click in the wrong position to select a photo and to close the picker. This basically makes it unusable.
To demonstrate, use Xcode 26 on macOS 26 to create a new iOS app project based on Swift/Storyboard. Then update ViewController.swift with the following code:
import UIKit
import PhotosUI
class ViewController: UIViewController {
override func viewDidLoad() {
super.viewDidLoad()
var cfg = UIButton.Configuration.plain()
cfg.title = "Photo Picker"
let button = UIButton(configuration: cfg, primaryAction: UIAction(handler: { _ in
self.showPicker()
}))
button.translatesAutoresizingMaskIntoConstraints = false
view.addSubview(button)
NSLayoutConstraint.activate([
button.centerXAnchor.constraint(equalTo: view.safeAreaLayoutGuide.centerXAnchor),
button.centerYAnchor.constraint(equalTo: view.safeAreaLayoutGuide.centerYAnchor),
])
}
private func showPicker() {
var config = PHPickerConfiguration()
config.selectionLimit = 10
config.selection = .ordered
let vc = PHPickerViewController(configuration: config)
vc.delegate = self
self.present(vc, animated: true)
}
}
extension ViewController: PHPickerViewControllerDelegate {
func picker(_ picker: PHPickerViewController, didFinishPicking results: [PHPickerResult]) {
print("Picked \(results.count) photos")
dismiss(animated: true)
}
}
Then go to the "Supported Destinations" section of the project target. Add a "Mac (Mac Catalyst)" destination. Then under the "Deployment Information" section, make sure the "Mac Catalyst Interface" setting is "Scaled to Match iPad".
Then build and run the app on a Mac (using the Mac Catalyst destination) with macOS 26.0.1. Make sure the Mac has a dozen or so pictures in the Photo Library to fully demonstrate the issue. When the app is run, a simple screen appears with one button in the middle. Click the button to bring up the PHPickerViewController. Now try to interact with the picker interface. Note that all pointer interactions are in the wrong place on the screen. This makes it nearly impossible to choose the correct photos and close the picker.
Quit the app. Select the project and go to the General tab. In the "Deployment Info" change the “Mac Catalyst Interface” setting to “Optimize for Mac” and run the app again. Now the photo picker works just fine.
If you run the app on a Mac running macOS 15 then the photo picker works just fine with either “Mac Catalyst Interface” setting.
The problem only happens under macOS 26.0 (I do not have macOS 26.1 beta to test) when the “Mac Catalyst Interface” setting is set to “Scaled to Match iPad”. This is critical for my app. I cannot use “Optimize for Mac”. There are far too many issues with that setting (I use UIStepper and UIPickerView to start). So it is critical to the usability of my app under macOS 26 that this issue be resolved.
It is expected that PHPickerViewController responds correctly to pointer events on macOS 26 when running a Mac Catalyst app set to “Scaled to Match iPad”.
A version of this has been filed as FB20503207
My app uses the SwiftUI Map control to display annotations. When annotations contain buttons AND the map has an .onTapGesture modifier, annotation button taps aren’t always recognized.
Given the following ContentView, run it on an iOS device or simulator. Tap the buttons. Since iOS 18.0, some of the buttons won't respond to tap. I've also tried using the onTapGesture instead of a button, and that shows the same issue.
This was not a problem under iOS 17.x: it has only shown up for my users since updating to iOS 18. Additionally, this issue does not occur when running on either macOS 15.0 or visionOS 2.0 when running in "designed for iPad" mode.
Note that there was previously a tap issue on VisionOS 1.x (designed for iPad), where no tap gestures were received by annotations. I'm curious if this issue could be related.
I have filed a report in Feedback Assistant (FB15273909).
struct ContentView: View {
private let center = CLLocationCoordinate2D(latitude: 37.77925, longitude: -122.41924)
@State private var label: String = "tap a button"
@State private var locations: [Location] = []
var body: some View {
Map {
ForEach(locations) { location in
Annotation(location.name, coordinate: location.coordinate) {
Button(location.name) {
print("\(location.name) tapped")
label = "\(location.name) tapped"
}
.buttonStyle(.borderedProminent)
}
.annotationTitles(.hidden)
}
}
.onTapGesture { point in
print("Map tapped")
label = "map tapped"
}
.safeAreaInset(edge: .top) {
Text(label)
.padding()
.background(.thinMaterial)
.clipShape(.rect(cornerRadius: 10))
}
.navigationTitle("Test Page")
.navigationBarTitleDisplayMode(.inline)
.task {
for index in 1...16 {
locations.append(Location(name: "location \(index)",
coordinate: CLLocationCoordinate2D(latitude: center.latitude + Double.random(in: -0.02...0.02),
longitude: center.longitude + Double.random(in: -0.02...0.02))))
}
}
}
private struct Location: Identifiable {
let id: UUID = UUID()
let name: String
let coordinate: CLLocationCoordinate2D
}
}
#Preview {
ContentView()
}
We have separated much of our UI into different packages to reduce complexity and compile time. When we recently tested using new .xcstrings string catalogs, we hit an unexpected problem.
Strings extracted from SwiftUI components like Text or Button are extracted into the Localizable.xcstrings in the same package, but the default behaviour of Text(_ key:tableName:bundle:comment:) is to use Bundle.main.
When the default behaviour of the string extraction isn't to extract to the main app target, this introduces a very fragile system where it's easy to add code that looks localised, but ends up failing lookup at runtime.
I don't feel comfortable that we will always remember to define the correct module every time we create a Text. Also, other components like Button doesn't have an init that takes a Bundle, so we would also have to remember that Button(_ titleKey:action:) can now only be used in a package if we make sure that the main bundle contains a matching key.
Is there a way for us to make sure that strings are always extracted to the same place as they are resolved against by default? Either by having strings in packages extracted to an xcstrings file in the main app or having Text default to resolving against the module bundle by default?
We're building a UGC AR app and are leveraging App Clips to distribute AR experiences without app download.
Since earlier this week, many of our users are reporting sharing experiences as App Clip doesn't work anymore. They are getting the message "AppClip unavailable" on a little card. We attached a QR code to try it yourself and a link to a different experience. We tried with multiple experiences and on multiple devices already.
https://scenery.app/experience/1C925FDE-E49A-489B-BA14-58A4E532E645
Interestingly, we can't pinpoint the issue to an exact device or OS.
We tested on many devices and on most, the AppClip is being displayed as unavailable, stating "App Clip unavailable", whereas it works on a few. It all worked fine last week (before September 12th).
iPhone 13 Pro Max, iOS26: works
iPhone SE, iOS 17: works
iPhone 16 Pro, iOS 26: doesn't work
iPhone 12 Pro Max, iOS 26: doesn't work
iPhone 12 mini, iOS 18: does not work
iPad 9th gen, iOS 26: doesn't work
Please help. Our users are very dissatisfied as they expect this to work and it's a crucial feature.
We already filed a radar via Feedback assistant:
FB20303890
We have some rather old code that worked for many years, but recently started to crash sporadically here:
The crash looks like this:
or
Our code is called from many threads concurrently, but as said it worked without any issues until recently.
I've found the apparently same crash on iOS at this Reddit post: https://www.reddit.com/r/iOSProgramming/comments/1kle4h4/ios_185_doesnt_fix_cicontext_rendering_crash/
Recap: we believe it started on macOS 18.4 and is still on macOS 18.5. But maybe it was already on macOS 18.3. This matches the observation in the Reddit post well.
Should we create a feedback with sysdiagnose?
Thanks! :)
Hello,
creating a simple-as-it-gets Slider in SwiftUI and then running that app on Mac Catalyst with the macOS idiom enabled, the app crashes:
struct ContentView: View {
@State private var sliderValue: Double = 0.4
var body: some View {
VStack {
Slider(value: $sliderValue)
}
.padding()
}
}
running this will result in an exception:
_setMinimumEnabledValue: is not supported on UISlider when running Catalyst apps in the Mac idiom. See UIBehavioralStyle for possible alternatives.
This is obviously not ideal and also apparently no documented.
Is there a workaround for this?
It used to work for on macOS Sonoma.
macOS 26 RC
Xcode 26 RC
FB20191635
Thanks!
Basically when showing a view using the .fullScreenCover modifier, it has no background anymore, any other UI elements are still shown but the view under it is also still shown.
I'm building an iPad app targeting iPadOS 26 using SwiftUI. Previously, I added a custom button by overlaying it in the top-left corner:
content
.overlay(alignment: .topLeading) {
Button("Action") {
// ...
}
This worked until iPadOS 26 introduced new window controls (minimize/close) in that corner, which now overlap my button.
In the WWDC Session Video https://developer.apple.com/videos/play/wwdc2025/208/?time=298, they show adapting via .toolbar, but using .toolbar forces me to embed my view in a NavigationStack, which I don’t want. I really only want to add this single button, without converting the whole view structure.
Constraints:
No use of .toolbar (as it compels a NavigationStack).
Keep existing layout—just one overlayed button.
Support automatic adjustment for the new window controls across all window positions and split-screen configurations.
What I’m looking for:
A way to detect or read the system′s new window control safe area or layout region dynamically on iPadOS 26.
Use that to offset my custom button—without adopting .toolbar.
Preferably SwiftUI-only, no heavy view hierarchy changes.
Is there a recommended API or SwiftUI technique to obtain the new control’s safe area (similar to a custom safeAreaInset for window controls) so I can reposition my overlayed button accordingly—without converting to NavigationStack or using .toolbar?
Hello, community and Apple engineers. I need your help.
Our app has the following issue: NavigationStack pushes a view twice if the NavigationStack is inside TabView and NavigationStack uses a navigation path of custom Hashable elements.
Our app works with issues in Xcode 18 Beta 13 + iOS 18.0. The same issue happened on previous beta versions of Xcode 18.
The issue isn’t represented in iOS 17.x and everything worked well before iOS 18.0 beta releases.
I was able to represent the same issue in a clear project with two simple views. I will paste the code below.
Several notes:
We use a centralised routing system in our app where all possible routes for navigation path are implemented in a View extension called withAppRouter().
We have a enum RouterDestination that contains all possible routes and is resolved in withAppRouter() extension.
We use Router class that contains @Published var path: [RouterDestination] = [] and this @Published property is bound to NavigationStack. In the real app, we need to have an access to this path property for programmatic navigation purposes.
Our app uses @ObservableObject / @StateObject approach.
import SwiftUI
struct ContentView: View {
@StateObject private var router = Router()
var body: some View {
TabView {
NavigationStack(path: $router.path) {
NavigationLink(value: RouterDestination.next, label: {
Label("Next", systemImage: "plus.circle.fill")
})
.withAppRouter()
}
}
}
}
enum RouterDestination: Hashable {
case next
}
struct SecondView: View {
var body: some View {
Text("Screen 2")
}
}
class Router: ObservableObject {
@Published var path: [RouterDestination] = []
}
extension View {
func withAppRouter() -> some View {
navigationDestination(for: RouterDestination.self) { destination in
switch destination {
case .next:
return SecondView()
}
}
}
}
Below you can see the GIF with the issue:
What I tried to do:
Use iOS 17+ @Observable approach. It didn’t help.
Using @State var path: [RouterDestination] = [] directly inside View seems to help. But it is not what we want as we need this property to be @Published and located inside Router class where we can get an access to it, and use for programmatic navigation if needed.
I ask Apple engineers to help with that, please, and if it is a bug of iOS 18 beta, then please fix it in the next versions of iOS 18.0
I’m not seeing Liquid Glass on any standard components. A month ago around July 17th I ran our app and saw Liquid Glass on our tab view and various standard components. Those components have not been changed and yet I’m no longer seeing Liquid Glass in our app at all.
Components that were previously liquid glass but now are not include TabView and back navigation buttons.
I set the UIDesignRequiresCompatibility key explicitly to false but no luck. I was seeing this in Beta 7 and Beta 8 on a real device and on a sim.
I noticing that Monterey defaults to the NSWindowToolbarStyleAutomatic / NSWindowToolbarStyleUnified toolbar style, which suppresses the "use Small Size" menu item and customization checkbox.
So I've set the window to use NSWindowToolbarStyleExpanded. However, the toolbar will no longer change to a smaller icon size, as it did in MacOS 10.14, 10.15, and 11.0.
I've tried to set the toolbar item sizing to "Automatic" for all of our toolbar icons, but that results in bad positioning in both Regular and Small Size mode -- the height is way too big.
The native size of the icon .png files are 128 x 128. What's odd is that if I resize the window with the toolbar to be wider, the NSToolbarItems in the overflow area will be displayed in the toolbar are 128 x 128, where the rest of the toolbar icons get displayed as a 32 x 32 icon.
The only way to get it to layout remotely correct is to make the NSToolbarItem to have an explicit minimum size of 24 x 24 and maximum size of 32 x 32. And that USED to allow "small size", but on Monterey, it no longer does.
Anyone had any success with small size icons on Monterey?
[Also submitted as FB19313064]
The .disabled() modifier doesn't visually disable buttons inside a ToolbarItem container on iOS 26.0 (23A5297i) devices. The button looks enabled, but tapping it doesn't trigger the action.
When deployment target is lowered to iOS 18 and deployed to an iOS 18 device, it works correctly. It still fails on an iOS 26 device, even with an iOS 18-targeted build.
This occurs in both the Simulator and on a physical device.
Screen Recording
Code
struct ContentView: View {
@State private var isButtonDisabled = false
private var osTitle: String {
let version = ProcessInfo.processInfo.operatingSystemVersion
return "iOS \(version.majorVersion)"
}
var body: some View {
NavigationStack {
VStack {
Button("Body Button") {
print("Body button tapped")
}
.buttonStyle(.borderedProminent)
.disabled(isButtonDisabled)
Toggle("Disable buttons", isOn: $isButtonDisabled)
Spacer()
}
.padding()
.navigationTitle("Device: \(osTitle)")
.navigationBarTitleDisplayMode(.large)
.toolbar {
ToolbarItem {
Button("Toolbar") {
print("Toolbar button tapped")
}
.buttonStyle(.borderedProminent)
.disabled(isButtonDisabled)
}
}
}
}
}
When testing with iOS 18.4 Beta on iPhones which support Dynamic Island, after doing a Face ID authentication, the amount of time it takes before the AppDelegate method applicationDidBecomeActive() is called takes longer than iPhones that do not support Dynamic Island. The time it takes is about double, 1.2 seconds vs 2.5 seconds on average. This does not occur with versions before 18.4 Beta.
Anyone else seeing this?
Topic:
UI Frameworks
SubTopic:
UIKit
We're observing new crashes specifically on iOS 18.4 devices with this pattern:
Exception Type: SIGTRAP
Exception Codes: fault addr: 0x000000019bc0f088
Crashed Thread: 0
Thread 0
0 libsystem_malloc.dylib _xzm_xzone_malloc_from_tiny_chunk.cold.1 + 36
1 libsystem_malloc.dylib __xzm_xzone_malloc_from_tiny_chunk + 612
2 libsystem_malloc.dylib __xzm_xzone_find_and_malloc_from_tiny_chunk + 112
3 libsystem_malloc.dylib __xzm_xzone_malloc_tiny_outlined + 312
4 CoreGraphics CG::Path::Path(CG::Path const&) + 132
5 CoreGraphics _CGPathCreateMutableCopyByTransformingPath + 112
6 CoreGraphics _CGFontCreateGlyphPath + 144
7 CoreGraphics _CGGlyphBuilderLockBitmaps + 1112
8 CoreGraphics _render_glyphs + 292
9 CoreGraphics _draw_glyph_bitmaps + 1116
10 CoreGraphics _ripc_DrawGlyphs + 1464
11 CoreGraphics CG::DisplayList::executeEntries(std::__1::__wrap_iter<std::__1::shared_ptr<CG::DisplayListEntry const>*>, std::__1::__wrap_iter<std::__1::shared_ptr<CG::DisplayListEntry const>*>, CGContextDelegate*, CGRenderingState*, CGGStack*, CGRect const*, __CFDictionary const*, bool) + 1328
12 CoreGraphics _CGDisplayListDrawInContextDelegate + 340
13 QuartzCore _CABackingStoreUpdate_ + 612
14 QuartzCore ____ZN2CA5Layer8display_Ev_block_invoke + 120
15 QuartzCore -[CALayer _display] + 1512
16 QuartzCore CA::Layer::layout_and_display_if_needed(CA::Transaction*) + 420
17 QuartzCore CA::Context::commit_transaction(CA::Transaction*, double, double*) + 476
18 QuartzCore CA::Transaction::commit() + 644
19 UIKitCore ___34-[UIApplication _firstCommitBlock]_block_invoke_2 + 36
20 CoreFoundation ___CFRUNLOOP_IS_CALLING_OUT_TO_A_BLOCK__ + 28
21 CoreFoundation ___CFRunLoopDoBlocks + 352
22 CoreFoundation ___CFRunLoopRun + 868
23 CoreFoundation _CFRunLoopRunSpecific + 572
24 GraphicsServices _GSEventRunModal + 168
25 UIKitCore -[UIApplication _run] + 816
26 UIKitCore _UIApplicationMain + 336
27 app _main + 132
28 dyld __dyld_process_info_create + 33284
Key Observations:
Crash occurs during font glyph path creation (CGFontCreateGlyphPath)
Involves memory allocation in malloc's xzone implementation
100% reproducible on iOS 18.4, not seen in prior OS versions
Occurs during standard CALayer rendering operations
Not tied to any specific font family or glyph content
Questions for Apple:
Is this crash signature recognized as a known issue in iOS 18.4's CoreGraphics?
Could changes to xzone memory management in iOS 18.4 interact poorly with font rendering?
Are there specific conditions that might trigger SIGTRAP in CGPathCreateMutableCopyByTransformingPath?
Any recommended mitigations for text rendering while awaiting system updates?
Issue Summary:
On iOS 26.0.1 to 26.3, apps using multiple UITextFields for OTP input face a critical issue where the system autofill pastes the entire OTP string into a single text field, usually the focused one, rather than splitting digits across fields. Delegate events like textDidChange: do not trigger consistently on autofill, breaking existing input handling logic.
Expected Behavior:
OTP autofill should distribute each digit correctly across all OTP UITextFields.
Delegate or control events should fire on autofill to enable manual handling.
(BOOL)textField:(UITextField *)textField
shouldChangeCharactersInRange:(NSRange)range
replacementString:(NSString *)string {
if (string.length > 1) {
// Autofill detected - distribute OTP manually
for (int i = 0; i < string.length && i < self.arrayOTPText.count; i++) {
UITextField *field = self.arrayOTPText[i];
field.text = [NSString stringWithFormat:@"%c", [string characterAtIndex:i]];
}
UITextField *lastField = self.arrayOTPText[string.length - 1];
[lastField becomeFirstResponder];
return NO;
}
// Handle normal single character or deletion input here
return YES;
}
//
// Setup UITextFields - set .oneTimeCode on first field only
for (int i = 0; i < self.arrayOTPText.count; i++) {
UITextField *field = self.arrayOTPText[i];
field.delegate = self;
if (@available(iOS 12.0, *)) {
field.textContentType = (i == 0) ? UITextContentTypeOneTimeCode : UITextContentTypeNone;
}
}
What We’ve Tried:
Setting textContentType properly.
Handling OTP distribution in delegate method.
Verifying settings and keyboard use.
Testing on multiple iOS 26.x versions.
Impact:
Major usability degradation during OTP entry.
Forces fragile workarounds.
Inconsistent autofill reduces user confidence.
Request:
Request Apple fix OTP autofill to natively support multi-field UITextField OTP input or provide enhanced delegate callbacks for consistent behavior.
Did any one face this issue in recent time with iOS 26.0.1 to 26.3 beta version
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
It appears that hidesBottomBarWhenPushed no longer works in iOS 26 Beta 1.
Is it supposed to work, is it going away or is there a alternate behavior we should be using?
Topic:
UI Frameworks
SubTopic:
UIKit