A migration is probably the best solution but if you don't want to do that you could use a custom decoding for the enum extension Kind { init(from decoder: any Decoder) throws { let container = try decoder.singleValueContainer() let rawValue = try container.decode(String.self) if let kind = Kind(rawValue: rawValue) { self = kind } else { let oldValue = rawValue.capitalized // <-- You might need to adjust this depending on how the rest of the enum looks. guard let kind = Kind(rawValue: oldValue) else { throw DecodingError.dataCorrupted(DecodingError.Context(codingPath: [], debugDescription: "Unknow kind: \(oldValue)")) } self = kind } } } Note that this will fix the value when read from storage but to actually store the new value, "Credit", each objects will need to be saved until it's persisted.

If you create a Color.Resolved value in a playground and encodes it you see you get an array of doubles rather than a structure with RGB properties and the error clearly states that an array (keyless container) is not supported. So I guess the conclusion is that Color.Resolved isn't supported by SwiftData. My playground test code and output let color = Color.red var resolvedColor: Color.Resolved resolvedColor = color.resolve(in: EnvironmentValues()) let data = try! JSONEncoder().encode(resolvedColor) print(String(data: data, encoding: .utf8)!) [1,0.23137254,0.18823528,1]

If you create the query property using a FetchDescriptor then you can set a limit for the number of rows being fetched. The drawback of this solution is that it's not a one liner so you need to do it in the init @Query private var records:[Record] init() { var fetchDescriptor = FetchDescriptor<Record>(sortBy: [SortDescriptor(\Record.date, order: .reverse)]) fetchDescriptor.fetchLimit = 7 _categories = Query(fetchDescriptor) } If you for some reason don't want to do it in the init you could declare the fetch descriptor as a static variable and then pass it to the @Query declaration

ModelContext has a property autosaveEnabled that you can set to false and then handle validation and save logic in a button action or something

I am not sure exactly what you are asking but as far as the ModelContainer yo must supply all your model types, directly or indirectly, when you create the container object

But Hashtag and HashtagFilter are two different models so why should their persistentModelID values be equal? Your variable naming confused me.

You should pass the id (persistentModelID), as you do in your code above, across actor boundaries and never the object itself.

where does the content of filterHashtags come from?

If I understand you correctly the issue is that you can't use the word throws in your code as an attribute name but maybe you can work around that by using backticks around it. I know this works fine for variables and properties and in custom types so hopefully it works within a Core Data/SwiftData context as well. So something like this should work, var `throws`: String

https://stackoverflow.com/a/79090346/9223839

I only use the @Relationship macro at one end of the relationship to avoid any issues. Maybe something has changed in the latest version that allows us to use it at both properties but in your case there is really no reason to do so.

Try to remove the database file so that SwiftData can generate a new one.

Looks like you need to perform a migration, see for example https://developer.apple.com/wwdc23/10195

Your model looks a bit strange to me. Why do you have a many to many relationship between the two models and why do you have an id property of type Float? Is it correct to assume that the float is the actual value (size) here and if so why is then the value also the identifier and why does each value need to be unique?

If you want to do something like this I believe you should have one ModelContainer but with multiple ModelConfiguration objects.