Question

I have recently started teaching myself C# and Asp.net. I am trying to build a simple blog application. I am confused about repository pattern usage. I have seen few tutorials and the implementation varies.

For my blog application, I have two database tables (models) – blogs and comments. Currently, I have a IDbContext which looks like this:

    public interface IDBContext
    {
        IQueryable<Blog> FindAllBlogs();
        IQueryable<Blog> FindBlogsInMonth(int month);
        Blog GetBlog(int id);
        void Add(Blog blog);
        void Update(Blog blog);
        void Delete(Blog blog);

        void Add(Comment comment);
        //void Remove(Comment comment);
    }

and I have repository which looks like this:

    public class BlogRepository : IDBContext
    {
        private BlogDb db = new BlogDb();

        public IQueryable<Blog> FindAllBlogs()
        {
            return db.Blogs.OrderByDescending(b => b.PublishDate);
        }

        public Blog GetBlog(int id)
        {
            return db.Blogs.Single(b => b.BlogID == id);
        }
...
}

The other implementation of repository pattern is something like this:

public interface IDbContext
{
    IQueryable<Blog> Blogs { get; }
    IQueryable<Comments> Comments { get; }
    int SaveChanges();
    T Attach<T>(T entity) where T : class;
    T Add<T>(T entity) where T : class;
    T Delete<T>(T entity) where T : class;
}

which calls a repository and there is separate class for queries.

What is the best method to do it?

Answer 1

The easiest method is to use Entity Framework directly, particularly the new DbContext functionality that showed up in Entity Framework 4.1 and later.

The context will contain DbSet properties – each is an implementation of the repository pattern that accomplishes all of the goals you described above. IDbSet can be used if unit testing support is required.

I’ve seen a lot of demos for the ASP.NET MVC repository patterns online that end up wrapping Entity Framework with a custom repository. It’s a waste of time – it’s code that wraps other code and doesn’t serve any direct purpose aside from adding needless complexity.