Example please? Say you've got a class Foo whose instances needs to do something when the app enters/leaves the background. So you add enterBackground and leaveBackground methods to Foo. Those need to be called from the appropriate app delegate methods. But the Foo instances spread around your app are members of other classes that are themselves members of other classes. In order to call Foo's enterBackground and leaveBackground methods, all of those parent classes need to have enterBackground and leaveBackground methods which are called by their parents and which call their children. If you add a Foo somewhere new, you have to add "plumbing" through all of its parent classes to the top of your object hierarchy so that its enterBackground and leaveBackground methods can be called. It's better not to do this, and to find some other way of connecting the Foo instances to the events. The NotificationCenter is an example of another way. It is not without its problems though; for example, unit testing is more difficult when a class has "back door" interfaces such as listening to global notifications.

how one is expected to do this I believe you're expected to not do it. This is not a viable business model for an app. (IMO.)

I doubt this is possible in any practical way. Have you seen any other apps that do this? Have you considered how to comply with money laundering regulations around the world?

Do you need the FAISS GPU functionality? I would imagine that getting its CPU code to work should be relatively easy. Combining Swift and C++ is easier than it used to be.

Over the years there have been quite a lot of posts here like yours. Few of the posters return to describe the final outcome. I do wonder what that means. Maybe they got reinstated and were too lazy to come back and tell us. Or maybe their forum accounts got terminated too. Probably the most important lesson is to not have a single source of income. Don't put all your eggs in one basket. One bit of advice from my own career: if a friend invites you to a party full of people you don't know, don't waste your time trying to chat up his sister. Seek out the other bored-looking men and express an interest in their boring jobs. Yes I'm talking about accountants and lawyers. Get their business cards. Then when someone tries to sue you for infringement, or Apple terminates your membership without warning, you can find the card of some really smart professional who can give you great advice - for free over a beer.

Both are fine. Sometimes, arranging the "plumbing" to get from your App Delegate to the place where the notification is needed can be tedious; you end up adding methods to a pile of intermediate objects to pass on the notification. (The "dependency injection" problem). In that case, using the notification center is easier.

How are you observing this communication? Can you observe the content of the packets? The delay between the outgoing packet and the response will give you a clue about how far away the recipient is!

Nothing has been announced; we don't know. You could look at previous deprecations to see if there is a pattern. Looking only at online services, i.e. not on-device SDK APIs, Apple have generally announced the removal dates straight away and they have been quit short periods (examples: IAP hosted content, XML App Store API). So it's actually unusual for them to announce a deprecation without a removal date. Are you calling verifyReceipt directly from your app, or from your server?

The verifyReceipt API was only ever useful from a server. Calling it from an app is very insecure. Why not compile OpenSSL? There is also all the StoreKit 2 stuff which I think will tell you all you need to know about the ISP on the device.

r32 t = 2.f * PI32 * (r32) soundState->runningSampleCount / framesPerCycle; r32 sineValue = sinf(t); …. soundState->runningSampleCount++; What is r32? Does runningSampleCount ever return to 0, or does it just keep increasing forever? I suspect that you are relying on sin(largenumber) == sin(largenumber mod 2pi), which is only approximately true. If runningSampleCount and framesPerCylce are both integers, try this: t = 2pi * (double)(runningSampleCount % framesPerCycle) / framesPerCycle

If you search the forum for nullptr_t you'll find some previous similar reports. They seemed to be due to peculiarities in the user's environment. Did you, perhaps, have some previous error and then set some environment variables to "fix" it ? Look for Eskimo's posts on how to debug. And BTW, "post the exact command" means copy and paste exactly what you did from the terminal into a post here, not writing "I just used CMakefile".

What exact command did you run?

Yesterday, half of my testers showed status of "-" and had lost the count of sessions. Now, that External Testers group has completely disappeared. In fact there is no longer an "External Testers" section in the left column of the screen. However I do have "All Testers" under "General Information" where I can see everyone, with the correct status and number of sessions etc. Possibly related: initially I didn't have "External Testers". A random stack-overflow post told me that "External Testers" only appears once you have created an Internal Testers group. So I created an empty internal testers group, then created an external testers group, then removed the internal group. I reported this as bug FB12769342, but Apple say it is the intended behaviour (*). That was all about three weeks ago. It was OK until today; now I can't see "External Testers" again. Maybe I shouldn't have deleted the empty internal testers group. Do others whose status is showing "-" find that the correct info is shown in "All Testers"? (*) I'm a one-person business. The idea of having "internal beta testers" using TestFlight before moving on to an external beta test may seem sensible if you're a business the size of Apple, but I'm sure it's not how most of us work!

I have a couple of apps that do something like this; you may even have seen them. If your experience with 3D graphics is limited then using a game engine (i.e. Unity) is likely to be a good choice. If you have specific questions about that (i.e. "I can't find a way how to use them with offline data") then asking on their mailing lists / forums is the thing to do. (My apps don't do that; I'm a do-it-all-myself-from-scratch sort of person, for better or worse.) One thing to worry about is numerical precision. What is your minimum feature size? Generally GPUs use single-precision floats. When you're modelling the whole world that can get a bit tight. Think about that before you go too far. Something else to think about is map projections. If you don't care about the polar regions, this isn't too difficult. If it's important that you get Antarctica right then some extra work is needed. If in doubt, I suggest using spherical geocentric 3D coordinates. DEM data compresses quite well with JXL; gdal may not support that yet though. You could store the whole globe on an iPad, at some medium resolution. More realistically, store the whole globe at low resolution and a few areas that you are interested in at a higher resolution. Good luck!

Half of my testers now show status of "-" and they have all lost the count of sessions.