I assume std::call_once() uses the same technique and therefore has the same excellent performance as dispatch_once. You can read what it does here: https://github.com/llvm/llvm-project/blob/main/libcxx/src/call_once.cpp There is a comment at the start about dispatch_once. It doesn’t do double-checked locking; it seems to unconditionally lock a pthread_mutex.

This is expected. You cannot write to the app bundle, it is read-only. Your mutable data needs to go elsewhere in the filesystem. There ought to be some introductory docs about this somewhere….

does paying app mean that you have to pay to acquire the game Yes or does it mean the game is free and generates revenue through ads? No, that's a free app.

I've read that updating the app doesn't overwrite the data of the current version, but instead places the new version of the app in a new directory entirely while the other, older version is then deleted. Where did you read that? You didn't ask Chat GPT, did you? Sounds like the sort of nonsense it would spout. It's not true. When your updated app runs, it sees the files that the previous version wrote in the Documents/ directory and elsewhere.

Assuming an individual, not business account: I don't believe that I had to supply proof of address when I registered, but that was a LONG time ago. I do recall, however, that when I changed my address (I moved house) I had to send proof of my new address. (And at that time, they wanted me to send the documents by fax! Ha!) So they were more concerned about the possibility of a fraudster taking over an existing, profitable, account than creating a new account with no income stream, which makes sense. They don't pay to Paypal, AFAIK. They need an IBAN.

Here is one scenario: user installs app on phone, app generates key, app stores key id in the filesystem. User gets a new phone and sets it up from a backup of the old phone; key id in filesystem is copied over. App runs on new phone, doesn’t know it’s on a different device, tries to use key id; corresponding key is not present; App Attest returns invalid key error. I believe there must be other scenarios in which this can happen though. I see quite large numbers of users who seem to have failed app attest. I don’t think they are all hacked devices.

Search the forum for “ld_classic”.

The 29th was a Friday, it's now Sunday. That's called the weekend. Be patient!

Consider compiling as objC++ and using std::call_once().

Do we know if this is something that can be fixed in an iOS 12.x+1 release, or in an XCode 15.x+1 release?

How exactly does it crash?

But until I hear something "official", I feel like I'm just guessing. This is primarily a developer-to-developer forum. I've given you an accurate reply one hour after you asked your question. Feel free to wait for Apple to answer, but you could be waiting a long time.

The response is supposed to be an email, and no it is not shown on the web page.

They want a screenshot showing what the app looks like after purchasing the IAP. Presumably there will be some difference somewhere, even if it’s just a list with an additional item in it.

Screenshots should not be created on device but with simulator. You're lucky if you can use the simulator for everything. Haven't we had this discussion before? Yes: https://developer.apple.com/forums/thread/728194 could try to tweak it in Preview application (resize image). I would not admit to "faking" screenshots in a public forum! Interestingly, App Store Connect seems happy to accept preview videos of the "wrong" size (i.e. recorded on an iPhone XS, 6.5-inch) in this case, but not screenshots.