Question

This issue may have be raised here before, but I am not confident that I have understood the solution. Here is the problem, in Objective-C, Core Data, for iOS.

I am going to model a twitter-like User-Following relationship, One user can follow many, and be followed by many. The Following itself has attributes such as createTime, pending, and level – that means, I cannot merely let User has bidirectional relationships like follower and followed, like :

User <<->> User

but add one more entity, Following, with relationships like following and followed.

User <->> Following <<-> User

Here are two questions:

1. Is this the best practice?

2. If so, how to write a [aUser valueForKeyPath: …]; to fetch his/her followers or friends (being followed).

Answer 1

Is this the best practice?

I don’t think this problem falls into the category of “situations for which a best practice has been established”, but it seems like a reasonable solution.

If so, how to write a [aUser valueForKeyPath: …]; to fetch his/her followers or friends (being followed).

Let’s say that Following has a follower property for the person doing the following, and a leader property for the person being followed. Also, User has a leads property for the Following relationships with people who follow a user and a follows property for the Following relationships with people that the user follows. Finally, User also has a name property.

Given all that, to get the names of aUser’s followers, you could say:

NSSet *followers = [aUser valueForKeyPath:@”leads.follower.name”];

and to get the names of people aUser follows:

NSSet *leaders = [aUser valueForKeyPath:@”follows.leader.name”];

It seems a little odd at first that you’d get sets back from those key paths since the last two keys in each path are singular. However, since the leads property returns a set of Following objects, leads.follower returns a set containing the users for each of those objects, and leads.follower.name returns a set containing the names of the users of those objects.