Post

Replies

Boosts

Views

Activity

Reply to [UITableViewDiffableDataSource] How to embed UITextView in UITableViewCell with UITableView.automaticDimension for cell height.
Hi! I have the same layout ( I guess ) where I put a textView ( actually, a subview of subview (of subview...) of cell content view is text view ). Sorry, and yes, it is only one solution so far. You can, for example, check this solution on iOS 14, maybe it is not required anymore and it works as expected ( automagically ). What about your question? Well, it is long story and it should be covered or solved easily in iOS 14 by Cell Configurations. Consider you have a ViewModel: class ViewModel { 	var first: Id 	var second: String } It is a simple model and it can be easily conformed to Hashable. But, however, we don't know how it will going in future, right? I suggest to add another creature: extension ListViewModel { struct Row { 	var viewModel: ViewModel 	var diffable: Diffable 	struct Diffable { 		var id: Id 		var first: String 		var second: Int 	} } } You add additional Row model which represents a model for your Cell and, especially, for a List of Cells. It is not the same as ListViewModel.Entry. It is an ItemIdentifierType for DiffableDataSource. I intentionally put it in ListViewModel namespace, because it works with ViewModel and embraces it into self and conforms to necessary protocols. extension ListViewModel.Row: Hashable { 	static func == (lhs: Self, rhs: Self) { 		lhs.diffable == rhs.diffable 	} 	func hash(_ inout hasher: Hasher) { 		hasher.hash(self.diffable) 	} } As you see, I separate an intention of Row and its nature. It is intended to be an ItemIdentifierType for DiffableDataSource, BUT, it is not what it actually is. Actually, it is projection of viewModel on properties of cell AND states of cell. So, here we distinguish a projection and ItemIdentifier. I called it Diffable, because we would check if our view states that are computed from model did change. Why this solution would work? First of all, it doesn't rely on view model itself, only on properties for cell and states for cell. Second, it requires that some structure ( Diffable in our case ) contains all these states and properties inside to compute difference. Does it look similar to Cell Configuration from iOS 14?
Jun ’20
Reply to Secure Transport is deprecated. Is there any replacement for SSLCreateContext function for C APIs?
Intention I would like to add adapter for networking framework to socket-based transport layer at libgit2 library. Library This library creates ssl context as /// TCP, right? st->ctx = SSLCreateContext(NULL, kSSLClientSide, kSSLStreamType); and sets security protocols as /// TLS SSLSetProtocolVersionMin(st->ctx, kTLSProtocol1) SSLSetProtocolVersionMax(st->ctx, kTLSProtocol12) libgit2 library defines an interface for socket-based api. You have to provide read/write functions for each "socket-based" adapter. The adapter write function signature is static ssize_t adapter_write(git_stream *stream, const char *data, size_t len, int flags) Adapters SecureTransport is relying on socket-based functions and it uses straightforward approach without callbacks. Read something, get result. Networking framework suggests a different approach with callbacks. So, instead of reading data in do-while loops, you have to add callbacks with "received/sent" partial result. Semaphore approach To adapt callback API I've added semaphore. Although I'm not sure this approach is efficient in terms of nw_connections. /// Rough draft static ssize_t apple_network_adapter_write(git_stream *stream, const char *data, size_t len, int flags) { apple_network_adapter_stream *st = (apple_network_adapter_stream *) stream; size_t data_len, processed; OSStatus ret; GIT_UNUSED(flags); data_len = min(len, SSIZE_MAX); nw_connection_t connection = ... ;/// retrieve connection dispatch_data_t ddata = dispatch_data_create(data, data_len, NULL, DISPATCH_DATA_DESTRUCTOR_DEFAULT); nw_content_context_t context = NW_CONNECTION_DEFAULT_MESSAGE_CONTEXT; /// We have to add semaphores for this API. /// Otherwise, it won't be able to "be" synced. dispatch_semaphore_t semaphore = dispatch_semaphore_create(0); __block int errno = 0; __block processed_length = -1; nw_connection_send(connection, ddata, context, true, ^(nw_error_t _Nullable error) { if (error == NULL) { processed = len; } else { errno = nw_error_get_error_code(error); } dispatch_semaphore_signal(semaphore); }); dispatch_semaphore_wait(semaphore, DISPATCH_TIME_FOREVER); ret = errno; if (ret != noErr) { return apple_network_adapter_error(ret); } GIT_ASSERT(processed < SSIZE_MAX); return (ssize_t)processed; } Hint Also I find another hint with it that nearly every object in networking framework is defined as NSObjectProtocol object.
Jun ’21
Reply to Integrating TouchID with Authorization Services
At Apple Labs engineer told me to use LAContext ( LocalAuthentication.framework ), because this object is adopting NSSecureCoding protocol and can be used to authenticate user. The flow is simple. Create LAContext at main GUI. Authenticate user via canEvaluatePolicy() and evaluatePolicy() Send LAContext through XPC to your desired service. Call canEvaluatePolicy() and evaluatePolicy() again. However, I have a problem with last part. I made a wrapper object which carries LAContext through XPC. ( AuthorizationPayload ). It works fine and carries LAContext to service from main GUI. However, when I try to verify LAContext on service side, I receive errors: LAContext[61791:0] failed to initialize: Error Domain=com.apple.LocalAuthentication Code=-10 "Context not found." UserInfo={NSDebugDescription=Context not found., NSLocalizedDescription=Authentication failure.} and canEvaluatePolicy() also returns false with the same error. Additional. I also added this sandbox rule to my service. (allow mach-lookup (global-name "com.apple.CoreAuthentication.daemon") ) What can I do in this situation?
Jul ’22