“build disappears”

Albert Pascual -- I am unable to send a code-level support request because I no longer have a paid developer account. Without going into grim detail, since I could no longer type very well after the accident, I had to leave my career as a developer. It sucked. I had been doing backend development back when there was a Mac OS X Server and an ADC membership was $500 a year. But being disabled as I am gave me a lot more appreciation for accessibility in macOS. Having these things working is important to my quality of life. Last night I installed 26.4 RC and there are multiple new issues. At this point I'm going to have to reset to 26.0 because too much of what I rely on either doesn't work or has become too irritating. A few examples: the orange dot in the upper right corner somehow became more distracting as it is now an orange square with a black border. Still just as useless but now extra annoying. Now when I say Hey Siri and my computer is locked, Siri responds with You will need to unlock you device first

The only way out of it was to create a build post-action script to generate the missing dsym file. Here is the gist of the script: set -euo pipefail if [ -z ${ARCHIVE_PATH:-} ]; then exit 0 fi APP_PATH=$ARCHIVE_PATH/Products/Applications/App_Product_Name.app if [ ! -d $APP_PATH/Contents ]; then exit 0 fi BIN=$APP_PATH/Contents/Frameworks/Framework_Name.framework/Framework_Name OUT=$ARCHIVE_PATH/dSYMs/Framework_Name.framework.dSYM OUT_DWARF=$OUT/Contents/Resources/DWARF/Framework_Name if [ ! -f $BIN ]; then exit 0 fi if [ ! -f $OUT_DWARF ]; then mkdir -p $ARCHIVE_PATH/dSYMs rm -rf $OUT xcrun dsymutil $BIN -o $OUT fi if [ ! -f $OUT_DWARF ]; then exit 1 fi BIN_UUIDS=$(xcrun dwarfdump --uuid $BIN | awk '{print $2}' | sort) DSYM_UUIDS=$(xcrun dwarfdump --uuid $OUT_DWARF | awk '{print $2}' | sort) if [ $BIN_UUIDS != $DSYM_UUIDS ]; then exit 1 fi

BTW, however I solve this, if I do, it needs to work on both Mac OS and iOS - I have no idea if there's much market for iOS Audio Units, but it was easy enough to build for both, so I will release both. It's at least likely to be a less saturated market :-)

Definitely a hell of an experience in the worse way. Apple Watch Series 4 iPhone 11 Mac Neo I had 0 issues developing/deploying to Pixel WearOS, but obstacles after obstacles, developing for and deploying to Apple WatchOS. After several days of working around each obstacle with the help of Gemini, I'm adding the summary of what I had to do so far to get to initial deployment in case this helps Apple/others with fix/workarounds: Here is the complete technical post-mortem for the development environment setup. This summarizes the specific hurdles cleared to get a Watch Series 4 (watchOS 10.6.2) communicating with your Mac via a No-SIM iPhone bridge. 🛠️ Environment Recovery Summary The Networking Layer The primary blocker was extreme network instability on the default 10.0.0.x gateway. The Symptom: ping tests showed 62% packet loss and 1,000ms+ latency. The Fix: Migrated all devices to a clean 192.168.86.x subnet. The Result: Achieved 0% packet loss and ~13ms latency, providing the stable pipe required for the

Thank you for the response! I think there may be a slight misunderstanding of my question. I am already linking each LeaderboardScore record to a user via a reference field. My challenge is how to efficiently calculate and determine the user’s rank, since I am not storing a rank field in the records and instead compute it client-side. Please see attached photo for visual reference. Current Implementation Fetch LeaderboardScore records sorted by score (descending): let query = CKQuery(recordType: LeaderboardScore, predicate: predicate) query.sortDescriptors = [NSSortDescriptor(key: score, ascending: false)] Assign rank based on position in the sorted results: entries = scores.enumerated().map { index, score in score.asLeaderboardEntry( rank: index + 1, // Rank assigned client-side isCurrentUser: score.profile.id == currentProfileID ) } This works well when the current user is within the fetched results (e.g., top 15–20). In that case, I can determine their rank directly from the local data. However, if the use

I am seeing a similar issue, but only for my Mac Catalyst archive on XCode 26. My app is a SwiftUI application that embeds an Objective-C framework built from a separate Xcode project. The framework binary itself does have UUIDs, but when I inspect the .xcarchive, the corresponding framework dSYM is missing from the archive’s dSYMs folder. What is confusing is that this does not happen for the iOS build. I use the same framework target, and when I archive the iOS app and upload to TestFlight, I do not get the warning. The problem only appears when I archive the Mac Catalyst version. So in summary: • The embedded framework binary is present in the archive • The framework executable has valid UUIDs • The app archive’s dSYMs folder does not contain the matching dSYM for that embedded framework • This only happens for Mac Catalyst • The same setup works correctly for iOS Has anyone seen Catalyst archives skip generating or packaging dSYMs for embedded frameworks, even though the binary itself contains de

Filed here: FB22276321 (Xcode Cloud build encountering persistent Connection Refused error)

> I have a custom hardware board that I want to communicate serially with from an iPad. Should I use the DriverKit route or the MFi route? Depends on exactly what your needs are and the hardware you're working with. Starting with MFi, the biggest advantage is that the software side (ExternalAccessory framework) is far simpler to use and includes support for background access (DriverKit does not). The main downside is that it requires a certain amount of custom hardware and the certification process. Comparing it with DriverKit is... complicated. The software side of DriverKit is FAR more complicated. To start with, the DriverKit API is a somewhat odd derivative of the IOKit kernel API. For someone who's familiar with IOKit, it's similar enough to be understandable, but for a new developer, the learning curve can be quite significant. None of this is helped by the fact that DriverKit is poorly documented[1], to the point that my standard advice to new developers is to study the IOKit kernel API (which has o

Thank you for reporting this — it's a clear and well-documented write-up. I was able to reproduce this on a MacBook M5 Max running macOS 26.3.1 (25D2128). I built a small test app that exercises both CGSetDisplayTransferByTable and CGSetDisplayTransferByFormula with several different gamma tables (identity, warm tint, inverted, red-only, gamma 1.8). On the M5 Max: Both APIs return kCGErrorSuccess Reading back the table with CGGetDisplayTransferByTable returns the correct values that were just set No visual change occurs on the display The same test app works correctly on an M5 (non-Max) machine running the same macOS build — all gamma changes are immediately visible on screen. This suggests the gamma tables are being stored correctly at the CoreGraphics level, but are not being applied to the display pipeline on M5 Max hardware. Tested on both built-in and external displays. I've related your two bugs (FB22273730 and FB22273782) together and attached the sample project I used to reproduce the problem

If you consistently see this, please file a bug report with a link to a build, and details of the URL that is unavailable during the build. Please post the FB number here for the record. — Ed Ford,  DTS Engineer