When I want legend inside the filedset, not on the border as default browsers design, I set float: left to the legend. It works in every modern browser - Safari too. The problem is when the filedset itself is set to display: flex. Float propertie of the legend is ignored. In other browsers this approach works fine and the float stays and the legend is inside of the fieldset as we expect.

Hello. Environment Safari version 18.1 (22B81). Problem In my Safari Web Extension i need to determine the url of the active tab. I do it this way: const onError = (error) => { console.error(error); }; const getFrameUrl = async (frames) => { console.info(frames) }; const getAllFrames = (frames) => { browser.webNavigation.getAllFrames({ tabId: frames[0].id }).then(getFrameUrl, onError); }; browser.tabs.onUpdated.addListener(() => { browser.tabs.query({ currentWindow: true, active: true }).then(getAllFrames, onError); }); The problem is that if the tab url is a link in the blob protocol, then the result function browser.webNavigation.getAllFrames({ tabId: tabs[0].id }) is as follows: {frameId: 0, parentFrameId: -1, url: "", errorOccurred: false} In Safari version 17 the correct URL value was returned: {frameId: 0, parentFrameId: -1, url: "blob://<some id>", errorOccurred: false} How can I fix the problem? Environment Safari version 18.1 (22B81).

Hi, iOS 18.1 Safari 18 does not play HLS/.m3u8 audio tracks (while all previous iOS Safari still do). If I turn on the Web Inspector toggle and put it on, restart Safari/ refresh the page, the player works. Before filing a bug, are you already aware of such behavior or some way to deal with it?

Starting Safari 18, Declarative Net Request redirections are no longer working (it used to be working on Safari 17). It seems to be related to the regex validation that is not working as expected. Here is an example of our DNR rule: { priority: 1, action: { type: 'redirect', redirect: { regexSubstitution: `${scheme}//${extensionHost}/index.html\\1#/\\2`, }, }, condition: { // app.dashlane.com, *.app.dashlane.com regexFilter: '^https?://w*\\.?app\\.dashlane\\.com(\\??[^/#]*)[^#]*#?/?(.*)$', resourceTypes: [ 'main_frame' ], }, } Is it a known issue ? Also, it seems to be related to this existing issue: https://forums.developer.apple.com/forums/thread/763505

We recently converted an existing app to adopt scenes (for CarPlay). The app has one main view controller that presents a WKWebView to show our content. Everything works fine in both landscape and portait mode when swiping on the screen to scroll. But with an iPad using an external Magic Keyboard, once you rotate to landscape mode the scrolling gestures are reversed. Swiping vertically on the trackpad is scrolling the page horizontally and vice versa. When this happens an error like below is logged (this error also shows up when in portait mode, but scrolling works as expected): Unexpected window orientation: <UIWindow: 0x10370d8f0; orientation: landscapeLeft (4)> { hidden = NO; frame = {{0, 0}, {1180, 820}}; bounds = {{0, 0}, {1180, 820}}; ownsOrientation = NO; ownsOrientationTransform = NO; autorotationDisabled = NO; windowInterfaceOrientation = unknown (0); rootTransformOrientation = landscapeLeft (4); viewTransformOrientation = unknown (0); autorotationDisabled = NO; orientationVC = ... { providedSupportedOrientations = ( Pu Ll Lr Pd ); resolvedSupportedOrientations = ( Pu Ll Lr Pd ); canPreferOrientation = NO; }; }, event type: 6 It seems to suggest that while the view controller orientation is set correctly. It doesn't know what the orientation is for the window. I have been unable to figure out what would change with adopting scenes that would explain this behavior. I assume it has to do with the potential multi-window nature of it but haven't found any docs that describe how to ensure the window is setup to use the same orientation as the device. Any suggestions on things to check?

The specification states that the timestamp property is a DOMHighResTimeStamp which always represents a value in milliseconds. However, in my testing (on iOS as I don't have a mac) the timestamp is given as seconds. This could break compatibility if an app/library is not tested in Safari. Worse is that the deviation seems to be undocumented as every compatibility chart states full compatibility and defines the property as milliseconds. Any workaround would also be up to one frame incorrect. So a stutter can affect input accuracy. I know this might seem small but it could mean that an app or game is unplayable in Safari if it relies on this for input.

When a video in Safari is played in fullscreen mode and the user pauses it, an overlay (such as a play button or other controls) appears. This affects the user experience, as we want to avoid showing the overlay during fullscreen viewing after the video is paused. Expected Behavior: The overlay should not appear after the video is paused in fullscreen mode. Steps to Reproduce: Open a webpage https://slides-dev.pitchavatar.com/rtxuf in Safari. Start playing the video. Enter fullscreen mode. Pause the video. The overlay appears after the video is paused. Screenshot: Additional Information: Safari version: Safari 18.0.1 (20619.1.26.31.7) Environment: macOS Sequoia 15.0.1 (24A348) We would like to know if there is there any way to prevent the overlay from appearing in Safari when the video is paused in fullscreen mode.

Description: When using a multilingual keyboard that includes Korean in iOS 18, the input element's onChange event is triggered multiple times instead of just once. This issue occurs not only when entering numbers with input type="tel" or inputMode="numeric", but also when entering text with input type="text". This causes unexpected behavior in forms and other input-related functionalities. Affected Devices and OS Version: Device: iPhone 16 Pro OS Version: iOS 18.0 You can reproduce the issue with this CodeSandbox example: https://codesandbox.io/p/sandbox/elegant-dream-jnqh39 Steps to reproduce: Use a multilingual keyboard (e.g., Korean and English) on iOS 18. Type some text into the input field (input type="text"). Also try entering numbers using input type="tel" or inputMode="numeric". Observe that the onChange event is fired multiple times for both text and numeric input. Expected behavior: The onChange event should only be triggered once when text or numeric input changes. Additional Information: This issue has been reported under feedback ID FB15377631. Currently waiting for a response from Apple regarding this feedback.

Hi Folks, do you know what happend with the "shape detection API" feature flag on Safari 18.X (IOs 18.X)?... in previous versions (17.X) i enabled the "shape detection API" feature flag and was able to detect codes like mentioned here... https://developer.mozilla.org/en-US/docs/Web/API/Barcode_Detection_API#browser_compatibility I built a PWA (Service Worker) with Angular 18 and facing this issue immediately after updating to IOS 18.0 (I enabled/disabled the flag, restartet the device several times... no success at all) Do you have an Idea what changed or how i can enable that feature again? Thx a lot in advance.. Cheers Martin

Hi! I have a rather complicated SwiftUI browser app with a WKWebView. There is an option to reload the website after a configurable amount of time. Starting with iOS 18, the app crashes after repeated reloads. How many reloads that are required depends on the device, sometimes 100, sometimes 1000. Reloading is done via a timer that triggers the following code on the main thread: let request = URLRequest(url: url, cachePolicy: policy) self.parent.webView.load(request) The URL is configurable and cachePolicy can be either .reloadIgnoringLocalAndRemoteCacheData or .useProtocolCachePolicy How the crash affects the device also differs from device to device and from time to time. I have suffered from the following crashtypes: App is killed App is killed and Safari also stops working App is killed and the whole OS is really slow The WKWebView stops loading and hangs at 20%. The device is rebooted My app has an option to disable cache. Cache is disabled by setting cachePolicy to .reloadIgnoringLocalAndRemoteCacheData and by removing all cache in a rather complicated way. Basicly i'm doing something like this: dataStore.removeData(ofTypes: types, modifiedSince: Date.distantPast, completionHandler: nil) if let klazz = NSClassFromString("Web" + "History"), let clazz = klazz as AnyObject as? NSObjectProtocol { if clazz.responds(to: Selector(("optional" + "Shared" + "History"))) { if let webHistory = clazz.perform(Selector(("optional" + "Shared" + "History"))) { let o = webHistory.takeUnretainedValue() _ = o.perform(Selector(("remove" + "All" + "Items"))) } } } if let cachesPath = NSSearchPathForDirectoriesInDomains(.cachesDirectory, .userDomainMask, true).first { let contents = (try? FileManager.default.contentsOfDirectory(atPath: cachesPath)) ?? [] for file in contents { if foldersToDelete.contains(file) { let path = cachesPath.appending("/").appending(file) do { try FileManager.default.removeItem(atPath: path) } catch { print("Can't delete cache file: \(path), error: \(error.localizedDescription)") } } } } The cache state affects the intensity of the crash. Disabling the cache shortens the time the app is working, while enabling the cache reduces the intensity of the bug. Based on my investigation, I suspect that loading a website in a WKWebVew leaks memory in iOS 18. If the whole website needs to be requested (= caching off), it results in a more significant memory leak and a faster crash time. Is this a known issue? Am I doing something wrong? Is there a potential workaround?

When I run the following code in XCode iOS Simulator on Safari (connecting via Safari DevTools): navigator.serviceWorker.ready.then((reg) => reg.pushManager.subscribe({ userVisibleOnly: true, applicationServiceKey: "..." }).then((sub) => console.log(sub.toJSON()))) I get the response: { keys: { p256dh: "", auth: "" } } But I'm expecting p256dh and auth (and endpoint) to be filled out. Why is it not?

In Safari Web Extensions on iOS 18, declarativeNetRequest Rulesets and Dynamic Rules take over twice as long to apply -- which causes rulesets to often fail to apply before sites load. In a boilerplate Xcode project you can note the time difference toggling the OISD (https://oisd.nl) ruleset on iOS 17 and iOS 18 simulators. Additionally, if you force quit Safari and reopen to a site with ads blocked by OISD list (e.g. espn.com) the content will be blocked in that initial state on iOS 17, but not in iOS 18 due to the latency. Based on the boilerplate extension, this bug is impacting all Mobile Safari Extensions using declarativeNetRequest Rulesets and Dynamic Rules. We know several other extension developers dealing with this issue. Our team wrote detailed reproduction steps in our Feedback Assistant ticket: https://feedbackassistant.apple.com/feedback/15196130 but have received no responses. I would attach a screen recording here but it won't allow me.

I am working Web to App flow for my product. Tolink to my app from web service, I define uri scheme from app and use it from my web site. But, when I try to use this scheme from mobile safari web, It sometimes failed with message that uri is invalid. That is weird It failed SOMETIMES not always Even I tried with same url. In my device, there is current version app. Why this happen? How can I resolve this issue?

This is problematic when opening links to other websites that require separate authentication. A typical example would be a Gmail web app, and then trying to open a link to a GitHub PR. In the 18.0 release notes, this is mentioned as Added support for opening links directly in web apps on macOS. (113034778 It'd be great if this could be customizable so that one could choose the old behavior.

Since iPadOS 18.x WKWebView seems to have a bug within its Fullscreen API (which can be enabled via WKPreferences.isElementFullscreenEnabled). This bug has the effect that websites trying to make an element (for example a video player) fullscreen fail to do so. This does not always happen, most of the time the fullscreen mode does work fine, but sometimes (far too often to be ignored) it does not. If an instance of WKWebView shows this issue, it can not be "fixed" by reloading the page or loading other pages, this issue exists in this instance forever. My App is a web browser App so I can create and remove WKWebView instance easily (by opening or closing Tabs). And there are times where I never see this bug, and times where ever other tab shows this bug. It's totally unreliable. The App does not show any issues at all when running under iPadOS 17 or older. The issue is only present under iPadOS 18.x. After some testing I've found out that when the bug has affected an instance of WKWebView, the JavaScript call element.requestFullscreen() will work if the element is a video element, but does no longer work if it is another element (like a DIV). If an instance of WKWebView is not affected by this bug, element.requestFullscreen() will work for all HTML elements. Does anyone has experienced this bug as well? And maybe found a workaround? Or maybe found a pattern which helps to find out what exactly is triggering this bug?

I'm developing an application that utilizes WKWebView to display a webpage. My goal is to improve the performance and loading speed of this webpage by intercepting specific API calls and providing pre-stored JSON responses directly from within the app, ensuring the webpage renders instantaneously. The challenge I am encountering is the inability to intercept HTTPS URL schemes directly from WKWebView. Any insights or solutions from the community on how to effectively manage and overwrite these particular network requests would be greatly appreciated.

When user opens a new tab and loads a website, the declarativeNetRequest.getMatchedRules() returns nil and on reloading in the same tab it returns the URLs blocked. Is this an expected behaviour?

Hi, I had been working on an extension without any issues, running it from Xcode and debugging locally in Safari. However, without making any code changes, and after uninstalling the extension, it stopped appearing in Safari after building for local debugging. Does anyone know what might be happening or how I can clean up the system? Thanks a lot and best regards.

Hi. I've noticed on both iOS 17 and iPadOS 16 that when an user opens a website in a WKWebview (e.g from Gmail), then requests it to be opened in the external Safari app (through the "compass" icon), their localStorage data is not transferred to the Safari tab. This behaviour breaks web experiences that rely on data being locally stored, e.g auth tokens or user identification data. It effectively stops users from being able to use some websites outside of an in-app context. I am aware of Webkit's tracking prevention mechanisms such as https://webkit.org/tracking-prevention/#partitioned-third-party-localstorage, but I don't think this should apply to this case. Here the user is navigating between two Safari tabs (one internal and one external) on the same domain, exact same URL, by pressing a native Safari webview icon. There is no third-party cross-domain tracking happening. Is this a bug or intended behaviour?

I have noticed that tabs.onRemoved works differently in Safari 18 comparing to other browsers and Safari 17. Open a tab e.g. apple.com Take note of the active tab id using this code: browser.tabs.query({currentWindow: true, active: true}, (x) => console.log(x[0].id)) Add a listener on onRemoved: browser.tabs.onRemoved.addListener(console.log) on the active tab, navigate to e.g. https://developer.mozilla.org Take note of the active tab id again (using the same tabs.query). Expect: there should be no console.log of onRemoved. the active tab id stays the same. Actual: there is a console.log of onRemoved. the active tab id is changed. Please help. If this is a bug introduced in Safari 18, it would break a lot of JS Web extensions.