existingObjectWithID:error: returns nil, but objectWithID: returns an actual usable object

Question

My understanding from the documentation and from this answer is that if the data exists, NSManagedObjectContext's existingObjectWithID:error: and objectWithID: methods should return the same object, but when the data doesn't exist, existingObjectWithID:error: will return nil while objectWithID: will return an object that has faults instead of data.

What I'm seeing in an application is an instance where (after creating the object on a background thread within a child managed object context and saving, then going to the main thread, saving, and bringing the object ID from the child context to the parent object context), existingObjectWithID:error: returns nil, but objectWithID: returns an actual usable object with valid data, not faults.

Is my understanding of the two methods incorrect? Am I doing something wrong?

(I want the returns-nil-when-there's-no-data behavior of existingObjectWithID:error:, but the inability to get the data for newly-created objects is problematic.)

---

edit: I suppose I could use objectWithID:, then immediately test accessing a property of the returned object within a try-catch block, catching the thrown exception, and replacing the faked object with nil (as is done here), but try-catch is expensive in Objective-C and this seems like a really bad idea.

Answer 1

The problem could be in temporary object IDs. Object ID is not permanent until it is saved to the store. So the question is when do you get the object ID from a managed object in child context: before you save the parent or after.

If you do this before you save the parent (which in turn, if parent is configured with persistent store coordinator and not with another parent, results in saving to the store), then you probably get temporary object ID. And for some reasons that are not disclosed to us by Apple one of the methods that return managed objects from object ID works, but the other doesn’t.