测试实体框架模型

Question

I'm going to test my EF Models. In order to do this I've create IDbContext class. But I don't know how to rewrite my Save and Delete methods, because I don't know how to write db.Partner.AddObject(obj); How to rewrite these methods?

public interface IDbContext
    {
        int SaveChanges();
        DbSet<Partner> Partner { get; set; }    
    }    
public class PartnerRepository : IPartnerRepository
{
    readonly IDbContext _context;
    public PartnerRepository()
    {
        _context = (IDbContext)new VostokPortalEntities();
    }
    public PartnerRepository(IDbContext context)
    {
        _context = context;
    }

    public void Save(Partner obj)
    {
        using (var db = new VostokPortalEntities())
        {
            if (obj.PartnerID == 0)
            {
                db.Partner.AddObject(obj);
            }
            else
            {
                db.Partner.Attach(obj);
                db.ObjectStateManager.ChangeObjectState(obj, System.Data.EntityState.Modified);
            }
            db.SaveChanges();
        }
    }
    public void Delete(Partner obj)
    {

        using (var db = new VostokPortalEntities())
        {

            db.Partner.Attach(obj);
            db.ObjectStateManager.ChangeObjectState(obj, System.Data.EntityState.Deleted);
            db.SaveChanges();
        }
    }
    public List<Partner> GetAll()
    {
        using (var db = new VostokPortalEntities())
        {
            return db.Partner.OrderByDescending(i => i.PartnerID).ToList();
        }
    }
}

Is this proper way to test EF Models?

Answer 1

Unit-testing of repositories takes a lot of time and does not give you many benefits. Why? Because repository don't have complex business logic. Usually there is pretty simple calls to underlying data-access API (ie ORM). I think it's match better to spend time on writing full-stack acceptance tests, which also will show if your repository do its job.

BTW there is interesting rule Don't Mock what you don't own :

By testing interactions with a mocked version of type we don't own, we really are not using our test to check for the correct behavior, nor to drive out a collaborator's design. All our test is doing is reiterating our guess as to how the other type works. Sure, it's better than no test, but not necessarily by much.