Thanks for the confirmation! I have read all your posts and I kind of knew it didn't make any sense, but it's very difficult to convince clients that something that they have been doing for years is not correct (especially when it passes notarization/gatekeeper checks).

@eskimo The number of iterations is actually configurable, but I wouldn't expect it to be too high (< 100). However, I would still like to explore if there are any other options.

@eskimo Closing and reopening the socket does work, but I don't know if it's a good idea to create a socket in every iteration. I also tried to set the IP_TTL option using ancillary data (similar to IP_TOS) but I guess that is not supported. Is there any other option (other than writing my own IP header)?

@eskimo The value of the socket option SO_NWRITE just after the sendto call is always 0. int value = 0; socklen_t len = sizeof(value); int res = getsockopt(sd, SOL_SOCKET, SO_NWRITE, &value, &len);

Apple-oriented developers tend to create a framework, and that has a single Info.plist file for all architectures. This issue can happen even if the user has a single Info.plist file for all architectures. In our case, a helper tool has a single Info.plist for both architectures with exactly the same keys and values but Xcode inserts a different value for the key BuildMachineOSBuild if the x86_64 and arm64 binaries are compiled on machines with different OS Build versions, resulting in verification failure.

NSXPCConnection has no idea how to serialise the name field. It does seem to be able to serialize the name field, but only the first struct in the array gets serialized. because NSXPCConnection has no idea how long that array is I was also not expecting it to work due to the same reason, but what does the documentation mean by C structures and arrays containing only the types listed above as acceptable parameters for methods and reply blocks? You have a few options here: For simple structs, I am using JSON to serialize/deserialize as I am not too familiar with Objective-C. Would any of these options help with function pointers? For more context, I am working on porting an application to Apple Silicon, but don't have the source code for some decade-old C libraries. So creating an XPC service that loads those libraries under Rosetta while the rest of the app runs natively on Apple Silicon. Some APIs in the library return some weird types involving pointers like a linked list where each node is itself a list of function pointers. I’d appreciate you filing a bug about that Filed FB9991036. To what end? Ideally for the lifetime of the application. The XPC service needs to run as long as the app is running and maintain state, the list of function pointers.

No. The observers are always added an removed from the same thread.

Can you share some documentation regarding this path? One concern with this approach is that other installers can overwrite some common library (eg libcurl.dylib, assuming we can also place dylibs here). And we do include an uninstaller.

Isn't this the correct way to launch the helper app? The main app launches the helper app when it needs to perform some background processing and it can outlive the main application.

Yes, it is an application bundle (.app) but does not have a GUI. It is launched by the main application using [[NSWorkspace sharedWorkspace] launchApplicationAtURL:....].

Why not embed those files within the app itself? Currently those files are within the app. They are basically html files that by default show the content bundled with the app. Users have the option to get extra content from the internet (banners, themes etc) that gets downloaded to the app's container. After upgrading to WKWebView, it seems both the html and resources need to be placed in the same directory hierarchy if the app is sandboxed, otherwise the webpage fails to load local resources. That is why I wanted to place those html files in the app's container after installation so that both the html and resources are in the same directory hierarchy.