Question

So, I have two tables, call them User and Item. I want each user to be able to flag (like “staring” things in the Google universe) any item.

Now, from a set-theoretic perspective it should be reasonable to store this as a simple relation between users and items. Call this new table UserItemFlags. The table should have two columns; one with foreign keys to the User table, the other with foreign keys to the Item table. If user U flags item I, this is then represented by the presence of the row (U, I) in the UserItemFlags table.

Now, my problem with expressing this in Fluent-NHibernate is this: I can’t seem to understand what to do with the Id mapping. Ideally, I would have no ID, because there is nothing more to it than presence or absence of this relation between a User and an Item. A natural consequence of this model is that it’s impossible to have duplicate rows in the table. This is a desired feature.

using FluentNHibernate.Mapping;

public class UserItemFlagsMapping : ClassMap<UserItemFlags>
{
    public UserItemFlagsMapping()
    {
        // Aaaa! I must have an Id!
        References(x => x.User).Not.Nullable();
        References(x => x.Item).Not.Nullable();
    }
}

I’d be happy for any pointers in the right direction, even if they would require me to abandon Fluent for this specific case.

Answer 1

What do you mean you don’t have an ID? You have a composite ID right there of both user ID and item ID.

---

Edit: The answer is in this comment:

The Id() method is only meant for single column primary keys. You want to use CompositeId() instead. Something like the following: CompositeId().KeyReference(x => x.User, “user_id”).KeyReference(x => x.Item, “item_id”; Obviously, if you’re using default convention mapping, you don’t need the column names.