widget configurationDisplayName and description Localization - swiftui

I want to localize the widget description.
I tried as below, but localization is not applied.
#main
struct RelaxOnWidget: Widget {
let kind: String = "RelaxOnWidget"
var body: some WidgetConfiguration {
StaticConfiguration(kind: kind, provider: Provider(), content: { entry in
RelaxOnWidgetEntryView(entry: entry)
})
.configurationDisplayName(LocalizedStringKey("Recently Played"))
.description(LocalizedStringKey("Quickly access recently played CDs."))
.supportedFamilies([.systemSmall])
}
}
It seems to support localization because it works as below, but I don't know why it doesn't. Can I get some advice on this?

Related

How to avoid rebuild view when tap on different button on NavigationSplitView

I have tried apple example
Bringing robust navigation structure to your SwiftUI app
so my code looks like this
NavigationSplitView(
columnVisibility: $navigationModel.columnVisibility
) {
List(
categories,
selection: $navigationModel.selectedCategory
) { category in
NavigationLink(category.localizedName, value: category)
}
.navigationTitle("Categories")
.toolbar {
ExperienceButton(isActive: $showExperiencePicker)
}
} detail: {
NavigationStack(path: $navigationModel.recipePath) {
RecipeGrid(category: navigationModel.selectedCategory)
}
}
Details View
struct RecipeGrid: View {
var category: Category?
var dataModel = DataModel.shared
var body: some View {
ZStack {
if let category = category {
ScrollView {
LazyVGrid(columns: columns) {
ForEach(dataModel.recipes(in: category)) { recipe in
NavigationLink(value: recipe) {
RecipeTile(recipe: recipe)
}
.buttonStyle(.plain)
}
}
.padding()
}
.navigationTitle(category.localizedName)
.navigationDestination(for: Recipe.self) { recipe in
RecipeDetail(recipe: recipe) { relatedRecipe in
NavigationLink(value: relatedRecipe) {
RecipeTile(recipe: relatedRecipe)
}
.buttonStyle(.plain)
}
}
} else {
Text("Choose a category")
.navigationTitle("")
}
}
}
var columns: [GridItem] {
[ GridItem(.adaptive(minimum: 240)) ]
}
}
My issue is if I go to details view then tap on other sidebar item
after that return to same tap, it will return to rootview also onAppear it toggled!
that mean the view did rebuild itself
on Apple News app it will stay on save view
it won't rebuild or return to rootview when I change sidebar item
I want same behavior, but I don't know how can I do it
I didn't find any question here or any article
explain how to do the same behavior as Apple News app
I think the solution is rather complex. I tried modifying the code myself but there’s quite a few things that need to change.
Firstly, the problem is caused because here NavigationStack(path: $navigationModel.recipePath) the recipePath gets reset every time the users taps the sidebar. I wish I understood why, but I passed to path a manual binding Binding.init(get:set:).
Secondly, the route is [Recipe], so when you select another category it makes sense that this route will be emptied. Otherwise, when the user taps the sidebar, the selected item from the previous category would still show.
So you would need the route to be a dictionary ([Category: [Recipe]]) and you need to hold the route for each category.
After I made this change, I saw in Apple Documentation that you can restore the route after the view has appeared, probably this solves the first problem.
So every time the category would change, I would make a copy of the route for that category. Then I tried to restore it when the category was tapped again, but I got stuck here because onAppear wasn’t triggered for me on RecipeGrid(…).onAppear {}.

How can I update/save a Property Change for Child Item from a Hierarchical List View of Items

See the following app screens:
Content View Screen:
Content View with hierarchical list children rows disclosed:
Parent Row Detail View:
Child Row Detail View:
Referencing the above views, here are the steps I do and the resulting problem I’m trying to solve:
Launch the app.
From the Functions (Content View) presented at launch, see that there is one item listed in a list view (1.0 Move Vessel)
Click the yellow (my app accent color) disclosure arrow at the right of the list item.
Two subordinate child list rows appear under the parent list item, 1.1 Move Position and 1.2 Hold Position.
When I tap the parent item (1.0 Move Vessel) in the hierarchy list, I'm successfully able to navigate to a detail view for that tapped item.
Edit the description of the 1.0 Move Vessel item (defaults to test) of the tapped item properties in the detail view using a TextEditor view.
Click yellow Save button at top left of detail view. The app navigates back to the parent Functions (Content View).
Click on the parent 1.0 Move Vessel row again.
See that description was successfully saved and now displayed from the change made in Step 5 and 6.
Repeat steps 5 through 8 again for 1.1 Move Position list row.
See that the edit/change made to the description was not saved and the default test1 description is displayed instead (not what is wanted).
Repeat steps 5 through 8 again for 1.2 Hold Position list row.
See that the edit/change made to the description was not saved and the default test2 description is displayed instead (not what is wanted).
I think I may have a problem in my save code logic and I'm trying to investigate.
Here are the swift files for the Detail View, the View Model, and the Model (I’ve not included the content view code because that code is working ok with the detail view. Again, I think the problem is in my save button and function call code for updating the view model.
NOTE: sorry that I can’t seem to figure out how to get all the code for a file contiguous in the code view. I seem to have some closing braces that don’t appear in the code view. I think you can still follow the code.
struct FunctionDetailView: View {
#State var vesselFunction: VesselFunction
#State var vesselFunctionDescription: String
#Environment(\.presentationMode) var presentationMode
#EnvironmentObject var functionViewModel : FunctionViewModel
var body: some View {
NavigationView {
Form {
Text("Enter description below")
TextEditor(text: $vesselFunctionDescription)
.frame(height: 200)
.toolbar {
Button {
//print(vesselFunction)
vesselFunction.funcDescription = vesselFunctionDescription
//print(vesselFunction)
functionViewModel.updateVesselFunction(vesselFunction: vesselFunction)
//print(vesselFunction)
presentationMode.wrappedValue.dismiss()
} label: {
Text("Save")
}
}
}
.padding()
.navigationTitle(vesselFunction.name)
.navigationBarTitleDisplayMode(.inline)
}
}
}
struct FunctionDetailView_Previews: PreviewProvider {
static var previews: some View {
FunctionDetailView(vesselFunction: VesselFunction(id: UUID(), name: "x.x Verb Noun", funcDescription: "Description", children: nil), vesselFunctionDescription: "placeholder")
.environmentObject(FunctionViewModel())
.preferredColorScheme(.dark)
}
}
FunctionViewModel.swift
#MainActor class FunctionViewModel: ObservableObject {
#Published private(set) var decomp : [VesselFunction] = [
VesselFunction(id: UUID(), name: "1.0 Move Vessel", funcDescription: "test", children: [
VesselFunction(id: UUID(), name: "1.1 Move Position", funcDescription: "test1", children: nil),
VesselFunction(id: UUID(), name: "1.2 Hold Position", funcDescription: "test2", children: nil)
])
]
func updateVesselFunction(vesselFunction: VesselFunction) {
/*
if let index = decomp.firstIndex(where: { (existingVesselFunction) -> Bool in
return existingVesselFunction.id == vesselFunction.id
}) {
//run this code
}
*/
// cleaner version of above
if let index = decomp.firstIndex(where: { $0.id == vesselFunction.id }) {
decomp[index] = vesselFunction.updateCompletion()
}
/*
else {
for item in decomp {
if item.children != nil {
if let index = item.children?.firstIndex(where: { $0.id == vesselFunction.id }) {
item.children![index] = vesselFunction.updateCompletion()
}
}
}
} */
}
}
FunctionModel.swift
struct VesselFunction: Identifiable {
let id : UUID
let name : String
var funcDescription : String
var children : [VesselFunction]?
init(id: UUID, name: String, funcDescription: String, children: [VesselFunction]?) {
self.id = id
self.name = name
self.funcDescription = funcDescription
self.children = children
}
func updateCompletion() -> VesselFunction {
return VesselFunction(id: id, name: name, funcDescription: funcDescription, children: children)
}
}
As you can see from the else and for-in loop code commented out at the bottom of the FunctionViewModel code, I was trying to see if I needed to do something like this code to access the children VesselFunction array entries of the decomp published property. With the if let index code that is not commented out, the save function works but only for the top-level decomp array VesselFunction elements, not the nested children arrays elements.
Any help would be appreciated so all decomp array elements, both parent and nested children, can be updated when the TextEditor field is changed and the Save button is pressed in the FunctionDetailView.
NOTE: I am only showing a 1 level deep nested array of children for the decomp property. I actually want to have multiple (at least 3) level of children arrays, so if you have any ideas how to make an updateVesselFunction function work for multiple children array elements, I would appreciate it.
In the main View use ForEach($model.items) { $item in so you get a write access to the model item. In the detail View change the #State to #Binding.
The issue isn't so much your code right now, as it is the architecture of the program. You really need to reorganize the app with MVVM concepts in mind. If you are not sure of them, study Apple’s SwiftUI Tutorials & Stanford’s CS193P. Without a proper architecture, you have gotten lost down a rabbit hole to the extent that I gave up trying to fix the code.
Also, given the structure of your data, I would give serious consideration to using Core Data to model it. Your VesselFunction struct contains an array of VesselFunction, and that it much better modeled as a relationship, rather than having a struct hold an array of the same struct which can hold an array of the same struct. It is a nightmare to deal with as a struct, instead of as a Core Data class.
I would also consider make your FunctionDetailView just display data, and have a separate editing view. This will keep your view separate and easier to manage.
Lastly, you have a lot of redundancy in your naming conventions. Theoretically, you could be trying to access a piece of data at functionViewModel.funcDescription (Not to mention: functionViewModel.children[index].children[subIndex].children[subSubIndex].funcDescription); this can get a bit unwieldy. The further you go down, the worse it will get.

SwiftUI view updates even if it does not depend on a dynamic property

I created an ObservableObject that gathers data and a view that depends on this object. More specifically some parts of the UI depend on one property of it and other parts depend on other properties.
The way ObservableObject works is that if any of its Published property gets updated (event if it does not change) it sends an objectWillChange notification that triggers updates for subscribed views.
However, I did expect that only the parts of the views that depends on an ObservableObject property that actually changed would be updated, not the entire view because body computations are expensive.
Unfortunately this it not the behavior I observed during my experiments. For instance in the following code the List view depends only on the color dynamic property of the ObservedObject "preferences" and the TextField updates only the text property of the observed object. I observed with the debugPrint's that when I type text the List and its rows gets updated each time even if it does not depend on text.
struct ContentView: View {
#State private var list = ["foo", "bar", "baz"]
#StateObject private var preferences = Preferences()
var body: some View {
VStack {
List(list, id: \.self) { element in
Text(element)
.foregroundColor(preferences.color)
.debugPrint("Row view updated")
}
Spacer()
HStack {
TextField("Text", text: $preferences.text)
.debugPrint("Text field view updated")
Button("Toogle Color", action: toggleColor)
.debugPrint("Button view updated")
}
}
.padding()
.debugPrint("Content view updated")
}
func toggleColor() {
preferences.color = [.blue, .green, .orange, .red].randomElement()!
}
}
final class Preferences: ObservableObject {
#Published var color: Color = .blue
#Published var text: String = ""
}
extension View {
func debugPrint(_ elements: Any...) -> Self {
#if DEBUG
print(elements)
return self
#else
return self
#endif
}
}
Is this the correct and intended behavior? How the optimal behavior I described above could be obtained, i.e. not calling body when the ObservedObject changes but only the components that depends on specific properties?
I observed in bigger SwiftUI projects that this behavior can greatly slow down the application and is not easily debuggable because unique state is often enforced through heavy ObservedObjects (or StateObject) injected at the root view and used in many places. I Observed this behavior even with small lists of components with heavy UI layout.
Note 1: I was bugged by the StateObject definition which suggests that only the views that depend on those properties are updated (and not when the StateObject changes).
SwiftUI creates a new instance of the object only once for each
instance of the structure that declares the object. When published
properties of the observable object change, SwiftUI updates the parts
of any view that depend on those properties [...]
Note 2: The issue totally disappear if I define the List in its own component, the rendering is fast:
struct ListView: View {
let list: [String] // Also works with a #Binding
let color: Color // Also works if it would be `preferences`
var body: some View {
List {
ForEach(list, id: \.self) { element in
Row(text: element, color: color)
}
}
}
}
Why in this case the view updates are fast?

SwiftUI View not updating on async change to published properties of Observed Object

I have the following SwiftUI View:
struct ProductView: View {
#ObservedObject var productViewModel: ProductViewModel
var body: some View {
VStack {
ZStack(alignment: .top) {
if(self.productViewModel.product != nil) {
URLImage(url: self.productViewModel.product!.imageurl, itemColor: self.productViewModel.selectedColor)
}
else {
Image("loading")
}
}
}
}
that observes a ProductViewModel
class ProductViewModel: ObservableObject {
#Published var selectedColor: UIColor = .white
#Published var product: Product?
private var cancellable: AnyCancellable!
init(productFuture: Future<Product, Never>) {
self.cancellable = productFuture.sink(receiveCompletion: { comp in
print(comp)
}, receiveValue: { product in
self.product = product
print(self.product) // this prints the expected product. The network call works just fine
})
}
The Product is a Swift struct that contains several string properties:
struct Product {
let id: String
let imageurl: String
let price: String
}
It is fetched from a remote API. The service that does the fetching returns a Combine future and passes it to the view model like so:
let productFuture = retrieveProduct(productID: "1")
let productVM = ProductViewModel(productFuture: productFuture)
let productView = ProductView(productViewModel: productViewModel)
func retrieveProduct(productID: String) -> Future<Product, Never>{
let future = Future<Product, Never> { promise in
// networking logic that fetches the remote product, once it finishes the success callback is invoked
promise(.success(product))
}
return future
}
For the sake of brevity, I've excluded the networking and error handling logic since it is irrelevant for the case at hand. To reproduce this as quickly as possible, just initialize a mock product with some dummy values and pass it to the success callback with a delay like this:
let mockproduct = Product(id: "1", imageurl: "https://exampleurl.com", price: "$10")
DispatchQueue.main.asyncAfter(deadline: .now() + 2.0, execute: {
promise(.success(mockproduct))
})
Once the product arrives over the network, it is assigned to the published product property.
The fetching works and the correct value is assigned to the published property. Obviously this happens after the view has been created since the network call takes some time. However, the View never updates even though the published object is changed.
When I pass the product directly through the View Model initializer rather than the future, it works as expected and the view displays the correct product.
Any suggestions on why the view does not react to changes in the state of the view model when it is updated asynchronously through the combine future?
EDIT: When I asked this question I had the ProductViewModel + ProductView nested inside another view. So basically the productview was only a part of a larger CategoryView. The CategoryViewmodel initialized both the ProductViewModel and the ProductView in a dedicated method:
func createProductView() -> AnyView {
let productVM = productViewModels[productIndex]
return AnyView(ProductView(productViewModel: productVM))
}
which was then called by the CategoryView on every update. I guess this got the Published variables in the nested ProductViewModel to not update correctly because the view hierarchy from CategoryView downwards got rebuilt on every update. Accordingly, the method createProductView got invoked on every new update, resulting in a completely new initialization of the ProductView + ProductViewModel.
Maybe someone with more experience with SwiftUI can comment on this.
Is it generally a bad idea to have nested observable objects in nested views or is there a way to make this work that is not an antipattern?
If not, how do you usually solve this problem when you have nested views that each have their own states?
I have been iterating on patterns like this to find what I think works best. Not sure what the problem is exactly. My intuition suggests that SwiftUI is having trouble making updates on the != nil part.
Here is the pattern that I have been using which has been working.
Define an enum for states in your networking logic
public enum NetworkingModelViewState {
case loading
case hasData
case noResults
case error
}
Add the enumeration as a variable on your "View Model"
class ProductViewModel: ObservableObject {
#Published public var state: NetworkingModelViewState = .loading
}
Update the state as you progress through your networking
self.cancellable = productFuture.sink(receiveCompletion: { comp in
print(comp)
}, receiveValue: { product in
self.product = product
self.state = NetworkingModelViewState.hasData
print(self.product) // this prints the expected product. The network call works just fine
})
Now make a decision in your SwiftUI based on the Enum value
if(self.productViewModel.state == NetworkingModelViewState.hasData) {
URLImage(url: self.productViewModel.product!.imageurl, itemColor: self.productViewModel.selectedColor)
}
else {
Image("loading")
}
Musings ~ It's hard to debug declarative frameworks. They are powerful and we should keep learning them but be aware of getting stuck. Moving too SwiftUI has forced me to really think about MVVM. My takeaway is that you really need to separate out every possible variable that controls your UI. You should not rely on checks outside of reading a variable. The Combine future pattern has a memory leak that Apple will fix next release. Also, you will be able to switch inside SwiftUI next release.

Simpler ViewModel implementation

I'm looking at an example of using SwiftUI with Combine: MVVM with Combine Tutorial for iOS at raywenderlich.com. A ViewModel implementation is given like this:
class WeeklyWeatherViewModel: ObservableObject, Identifiable {
// 2
#Published var city: String = ""
// 3
#Published var dataSource: [DailyWeatherRowViewModel] = []
private let weatherFetcher: WeatherFetchable
// 4
private var disposables = Set<AnyCancellable>()
init(weatherFetcher: WeatherFetchable) {
self.weatherFetcher = weatherFetcher
}
}
So, this makes some sense to me. In a view observing the model, an instance of the ViewModel is declared as an ObservedObject like this:
#ObservedObject var viewModel: WeeklyWeatherViewModel
And then it's possible to make use of the #Published properties in the model in the body definition of the View like this:
TextField("e.g. Cupertino", text: $viewModel.city)
In WeeklyWeatherViewModel Combine is used to take the city text, make a request on it, and turn this in to [DailyWeatherRowViewModel]. Up to here, everything is rosey and makes sense.
Where I become confused is that quite a lot of code is then used to:
Trigger a fetch when city is changed.
Keep hold of the AnyCancellable that's looking up weather data.
Copy the output of the weather look up in to dataSource by a sink on the weather fetch Publisher`
It looks like this:
// More in WeeklyWeatherViewModel
init(
weatherFetcher: WeatherFetchable,
scheduler: DispatchQueue = DispatchQueue(label: "WeatherViewModel")
) {
self.weatherFetcher = weatherFetcher
_ = $city
.dropFirst(1)
.debounce(for: .seconds(0.5), scheduler: scheduler)
.sink(receiveValue: fetchWeather(forCity:))
}
func fetchWeather(forCity city: String) {
weatherFetcher.weeklyWeatherForecast(forCity: city)
.map { response in
response.list.map(DailyWeatherRowViewModel.init)
}
.map(Array.removeDuplicates)
.receive(on: DispatchQueue.main)
.sink(
receiveCompletion: { [weak self] value in
guard let self = self else { return }
switch value {
case .failure:
self.dataSource = []
case .finished:
break
}
},
receiveValue: { [weak self] forecast in
guard let self = self else { return }
self.dataSource = forecast
})
.store(in: &disposables)
}
If I look in Combine for the definition of the #Published propertyWrapper, it seems like all does is provide projectedValue which is a Publisher, which makes it seem like it ought to be possible for WeeklyWeatherViewModel to simply provide the Publisher fetching weather data and for the view to make use of this directly. I don't see why the copying in to a dataSource is necessary.
Basically, what I'm expecting is there to be a way for SwiftUI to directly make use of a Publisher, and for me to be able to put that publisher externally from a View implementation so that I can inject it. But I've no idea what it is.
If this doesn't seem to make any sense, that figures, as I'm confused. Please let me know and I'll see if I can refine my explanation. Thanks!
I don't have a definitive answer to this and I didn't find a magic way to have SwiftUI make use of a Publisher directly – it's entirely possible that there is one that eludes me!
I have found a reasonably compact and flexible approach to achieving the desired result, though. It cut down the use of sink to a single occurrence that attaches to the input (#Published city in the original code), which substantially simplifies the cancelation work.
Here's a fairly generic model that has an #Published input attribute and a #Published output attribute (for which setting is private). It takes a transform as input, and this is used to transform the input publisher, and is then sinked in to the output publisher. The Cancelable of the sink is stored.
final class ObservablePublisher<Input, Output>: ObservableObject, Identifiable {
init(
initialInput: Input,
initialOutput: Output,
publisherTransform: #escaping (AnyPublisher<Input, Never>) -> AnyPublisher<Output, Never>)
{
input = initialInput
output = initialOutput
sinkCancelable =
publisherTransform($input.eraseToAnyPublisher())
.receive(on: DispatchQueue.main)
.sink(receiveValue: { self.output = $0 })
}
#Published var input: Input
#Published private(set) var output: Output
private var sinkCancelable: AnyCancellable? = nil
}
If you wanted a substantially less generic kind of model, you can see it's pretty easy to set up having the input (which is a publisher) be filtered in to the output.
In a view, you might declare an instance of the model and use it like this:
struct SimpleView: View {
#ObservedObject var model: ObservablePublisher<String, String>
var body: some View {
List {
Section {
// Here's the input to the model taken froma text field.
TextField("Give me some input", text: $model.input)
}
Section {
// Here's the models output which the model is getting from a passed Publisher.
Text(model.output)
}
}
.listStyle(GroupedListStyle())
}
}
And here's some silly setup of the view and its model taken from a "SceneDelegate.swift". The model just delays whatever is typed in for a bit.
let model = ObservablePublisher(initialInput: "Moo moo", initialOutput: []) { textPublisher in
return textPublisher
.delay(for: 1, scheduler: DispatchQueue.global())
.eraseToAnyPublisher()
}
let rootView = NavigationView {
AlbumSearchView(model: model)
}
I made the model generic on the Input and Output. I don't know if this will actually be useful in practice, but it seems like it might be.
I'm really new to this, and there might be some terrible flaws in this such as inefficiencies, memory leaks or retain cycles, race conditions, etc. I've not found them yet, though.
You can use URLSessionDataTaskPublisher and refactor out networking from all view models.
If you feel some part of the tutorial seems redundant, that is because it is.
MVVM in such usage is redundant and does not do the job better.
I have a refactored version (networking refactored, all view models removed) of that tutorial if you are interested in details.