Lambda42 Blog

Adventures with the Entity Framework and the Repository Pattern

2009-08-20T01:09:32+01:00

So far in my applications, I have been using the Entity Framework for object persistence (why? SQLite is why). Unfortunately, I also like to write lots of unit tests. Normally, this would be a problem, as the entity container would be coupled to the controllers.

How I implement the Repository Pattern

To get around this, I have been using the repository pattern to great effect.

I start by creating my IRepository<T> generic interface:

   1:      public interface IRepository<T>

   2:      {

   3:          T Single(Func<T, bool> filter);

   4:          IQueryable<T> All { get; }

5:

   6:          void Insert(T entity);

   7:          void Remove(T entity);

8:

   9:          void Commit();

  10:      }

People could say that the Single method is rather pointless, but I like to have it there. You could also comment on the lack of an Update method – but the entity framework’s change tracking does a good job. Other people would say this is bad, as the repository is being tied to the entity framework, but that is not the reason I am doing this.

Originally, I would have an interface per entity, but the generic interface solves all the duplication I was seeing.

This interface is extremely simple to implement, and solves my Entity Framework + TDD problems.

Regardless, roll on EF 4.0 POCOs!

Visual Studio 2010 jQuery intellisense

2009-06-13T00:14:19+01:00

Just a quick post for something I only today noticed in the Visual Studio 2010 beta – and its damned cool.

Yeah, that’s pretty awesome.

The problem with ASP.NET MVC - "spaghetti code"

2009-04-09T17:59:09+01:00

I apologise in advance if this post seems slightly incoherent - I write this whilst on a train, and its quite a bumpy ride. Also, another disclaimer - the code you're about to see will be soon heavily refactored and improved; don't expect it in this state to be amazing! For one, truncating the post content will be moved to the model level, and the post mark-up will be moved to a user control.

Consider the following mark-up:

<% foreach (Blog42.Model.Post p in Model.Posts) { %>

<%-- Post begin --%>

<div class="leftcontent">
    <h1><%= Html.ActionLink(p.Title,"Show", new { title = p.URI }) %></h1>
    <h2>Posted by <%= Html.ActionLink(p.User.UserName,"ByUser", new { username = p.User.UserName }) %>

     at <%= p.Date %></h2>
    
    <div class="blogcontent">
        <%= p.TruncatedContent %>
        <% if (p.IsTruncated) { %>
            <%= Html.ActionLink("Read More","Show", new { title = p.URI }) %>

        <% } %>
    </div>
                    
    <%= Html.RenderTags(p) %>
    
    <span><%= Html.ActionLink("Comments (" + p.Comments.Count + ")","Show", new { title = p.URI }) %></span>

</div>

<%-- Post end --%>

<% } %>

There are a couple of things here that I absolutely hate (and will soon in part be refactored) - the selection statement and the massive amount of <%= %>. For most people, they will see nothing wrong with this, and will keep doing it. In the end I believe this will lead to an un-maintainable application with unreadable mark-up.

Writing an helper is the solution to this - note that I have moved the tag building to a helper.

public static string RenderTags(this HtmlHelper helper, Post post)
{
    StringBuilder builder = new StringBuilder();

    if (post.Tags.Count >= 1)
    {
        builder.Append("<p><strong>Tags:</strong>");

        int tagNumber = 0;

        foreach (Tag tag in post.Tags)
        {
            tagNumber++;

            builder.Append(" ");
            builder.Append(helper.ActionLink(tag.Name,"ByTag","Post",new { tag = tag.Name }, null));

            if (tagNumber != post.Tags.Count)
                builder.Append(" | ");
        }

        builder.Append("</p>");
    }

    return builder.ToString();
}

This however, is also what I believe to be suboptimal - note the snippets of mark-up inside the code!

However, it is an improvement, and for the time being I believe a necessary evil. It does improve the readability of the mark-up, and introduces some semblance of reusability in the sense that I can use the same helper whenever I want to build a list of tags.

My experiences with ASP.NET Dynamic Data

2009-03-23T14:47:59Z

Not long ago I had no idea what the point of ASP.NET Dynamic Data was. I simply could not 'grok' it, until very recently when I tried using it for a client's CMS.

Basically, Dynamic Data turned this:

Into this:

Using only three lines of code in the Global.asax.cs:

MetaModel model = new MetaModel();

model.RegisterContext(typeof(BlogModel),

 new ContextConfiguration() { ScaffoldAllTables = true });

The site that has been generated for me lets my edit all my tables, filter by foreign keys, add new rows, view related rows, and more. All from three lines of code. Absolutely incredible.

ASP.NET MVC released

2009-03-20T14:12:03Z

Nice to see that finally ASP.NET MVC has been released! Been using it since the very first preview, and it's been a blast.

Can grab it here.

Also, noticed that Silverlight 3 has been released as a beta, already. This is a bit annoying as Silverlight 2 was only released 5 months ago. Too many technologies to master, so little time!

Consuming ADO.NET Data Services from jQuery

2009-03-10T18:23:28Z

One thing that has certainly sped up my development time is using ADO.NET Data Services with my AJAX. Instead of having to build controller actions or a web service to grab the data, i can just consume it from the data service directly.

I'm going to assume you know how to set up a Data Service - it really isn't too hard.

Actually getting jQuery to consume it isn't much harder. All you need is the standard '$.getJSON' function, and you're flying.

One thing that tripped me up, however, was the data that was returned from the service: a quick glance shows that the data is wrapped in a 'd' object.

{ "d" : [ (data here) ] }

ADO.NET Data Services

2009-02-12T18:25:23Z

Been playing around with ADO.NET Data Services this morning - really cool how you can point it at your (Entity Framework in this case) data model, and it fully exposes it via a web service.

Planning on using it to replace my current MVC JSON-returning controllers. Shouldn't be that much more difficult, it cleans up the controllers, and in the future will save masses amount of time. What's not to like?