Core Data with Swiftui Mvvm Feedback

Core Data with SwiftUI MVVM Feedback

For a cutting edge way to wrap the NSFetchedResultsController in SwiftUI compatible code you might want to take a look at AsyncStream.

However, @FetchRequest currently is implemented as a DynamicProperty so if you did that too it would allow access the managed object context from the @Environment in the update func which is called on the DynamicProperty before body is called on the View. You can use an @StateObject internally as the FRC delegate.

Be careful with MVVM because it uses objects where as SwiftUI is designed to work with value types to eliminate the kinds of consistency bugs you can get with objects. See the doc Choosing Between Structures and Classes. If you build an MVVM object layer on top of SwiftUI you risk reintroducing those bugs. You're better off using the View data struct as it's designed and leave MVVM for when coding legacy view controllers. But to be perfectly honest, if you learn the child view controller pattern and understand the responder chain then there is really no need for MVVM view model objects at all.

And FYI, when using Combine's ObservableObject we don't sink the pipeline or use cancellables. Instead, assign the output of the pipeline to an @Published. However, if you aren't using CombineLatest, then perhaps reconsider if you should really be using Combine at all.

How to pass/get Core Data context in SwiftUI MVVM ViewModel?

Here is your scenario

let contentView = MainView(context: context)          // << inject
        .environment(\.managedObjectContext, context)

struct MainView: View {
    @Environment(\.managedObjectContext) var context

    @ObservedObject private var viewModel: ViewModel // << declare

    init(context: NSManagedObjectContext) {
        self.viewModel = ViewModel(context: context)   // initialize
    }
}

Accessing Core Data Stack in MVVM application

You can use Core Data Singleton class

import CoreData

class CoreDataStack {
    static let shared = CoreDataStack()

    private init() {}

    var managedObjectContext: NSManagedObjectContext {
        return self.persistentContainer.viewContext
    }

    var workingContext: NSManagedObjectContext {
        let context = NSManagedObjectContext(concurrencyType: .privateQueueConcurrencyType)
        context.parent = self.managedObjectContext
        return context
    }

    // MARK: - Core Data stack

    lazy var persistentContainer: NSPersistentContainer = {
        let container = NSPersistentContainer(name: "MyStuff")
        container.loadPersistentStores(completionHandler: { storeDescription, error in
            if let error = error as NSError? {
                RaiseError.raise()
            }
        })
        return container
    }()

    // MARK: - Core Data Saving support

    func saveContext() {
        self.managedObjectContext.performAndWait {
            if self.managedObjectContext.hasChanges {
                do {
                    try self.managedObjectContext.save()
                    appPrint("Main context saved")
                } catch {
                    appPrint(error)
                    RaiseError.raise()
                }
            }
        }
    }

    func saveWorkingContext(context: NSManagedObjectContext) {
        do {
            try context.save()
            appPrint("Working context saved")
            saveContext()
        } catch (let error) {
            appPrint(error)
            RaiseError.raise()
        }
    }
}

Core Data is not thread safe. If you write something on manageObject and don't want to save that, but some other thread save the context, then the changes that you don't want to persist will also persist.

So to avoid this situation always create working context - which is private.

When you press save, then first private context get saved and after that you save main context.

In MVVM you should have DataLayer through which your ViewModel interact with Core Data singleton class.

How to preview Core Data objects in Canvas when using a CoreDataViewModel

The point of a preview is to use the ordinary view with some given data and not to duplicate the code from the actual view. Apart from that you seem to have much in place.

I would inject the Core Data manager class into the view model to start with:

class CoreDataViewModel: ObservableObject{        
    let manager: CoreDataManager
    
    @Published var cars: [Car] = []
    
    init(coreDataManager: CoreDataManager = .instance){
        self.manager = coreDataManager
        getCars()
    }

Then I would simplify the preview to

struct CarsView_Previews: PreviewProvider {
    static var previews: some View {
        CarsView(coreDataViewModel: CoreDataViewModel(coreDataManager: .preview))
    }
}

You are already creating one Car object in the preview so why not do that in a loop, here I have moved it into a separate function that can be called from inside the preview declaration or separately

#if DEBUG
extension CoreDataManager {
    func createMockCars() {
        for i in 1...5 {
             let car = Car(context: self.context)
             car.make = "Make \(i)"
             car.model = "Model \(i)"
        }
        try! self.context.save()
    }
}
#endif

As mentioned in one of the comments you should consider renaming your view model to something to do with cars like CarsViewModel or CarListViewModel.

---

Related Topics