LINQ to SQL - Save an entity without creating a new DataContext?

Question

I get this error

Cannot add an entity with a key that is already in use

when I try to save an Item

[AcceptVerbs(HttpVerbs.Post)]
public ActionResult Edit(Item item)
{
  Global.DataContext.Items.Attach(item);
  Global.DataContext.SubmitChanges();

  return View(item);
}

That's because I cannot attach the item to the global DataContext.

Is it possible to save an item without creating a new DataContext, and without having to assign each field of the item manually?

(I am very new to LINQ)

EDIT: I realised static DataContext would cause problems thanks to the comments below, it is now like this

public static AbcDataContext DataContext
{
  get
  {
    if (!HttpContext.Current.Items.Contains("DataContext"))
      HttpContext.Current.Items["DataContext"] = new AbcDataContext(ConnectionString);
    return (AbcDataContext)HttpContext.Current.Items["DataContext"];
  }
}

(Rex might not agree to that, but I can't be bothered changing the whole code at the moment - may be later)

Answer 1

Don't have a global/static DataContext, that is setting yourself up for pain. A DataContext should represent a single logical transaction ("get in, do x/y/z and get out"). They are cheap to create and easy to dispose; there is absolutely no reason to try to minimize them, much less keep a global/static one.

Answer 2

Suppose the primary key of your Item class is ItemId.

Suppose the ItemID for the instance you are attempting to update is 5.

The DataContext has seen an original state for ItemID 5, so it won't let you Attach(). http://msdn.microsoft.com/en-us/library/bb300517.aspx

In this version of Attach, the entity is assumed to be in its original value state. After calling this method, you can then update its fields, for example with additional data sent from the client.

There's three normal ways to perform an update in LinqToSql.

If the parameter to this Edit method was originally loaded up from the DataContext, then all you need to do is:

public ActionResult Edit(Item item) 
{
  Global.DataContext.SubmitChanges(); 
  return View(item); 
} 

The DataContext tracks changes against objects that it loaded. As a nasty side effect, any modified objects that was loaded by the DataContext are also going to be updated. This is a big reason to not use a single app level DataContext.

If the parameter to this Edit method was new'd up in your code, loaded by a different DataContext, or passed to your code (in other words, the instance has no attached DataContext) then you can do either of these:

public ActionResult Edit(Item item) 
{
  using(MyDataContext dc = new MyDataContext())
  {
//this new DataContext has never heard of my item, so I may Attach.
    dc.Items.Attach(item);
//this loads the database record in behind your changes
// to allow optimistic concurrency to work.
//if you turn off the optimistic concurrency in your item class
// then you won't have to do this
    dc.Refresh(item, KeepCurrentValues);
    dc.SubmitChanges(); 
  }
  return View(item); 
} 

public ActionResult Edit(Item item) 
{
  original = Global.DataContext.Items.Single(x => x.ItemID = item.ItemID)
  //play the changes against the original object.
  original.Property1 = item.Property1;
  original.Property2 = item.Property2;
  Global.DataContext.SubmitChanges(); 
  return View(item); 
} 

---

With your question answered, allow me to echo the concern that others have stated for using a static DataContext. This is a poor practice and goes against Microsoft's intended use of the DataContext class. http://msdn.microsoft.com/en-us/library/system.data.linq.datacontext.aspx

In general, a DataContext instance is designed to last for one "unit of work" however your application defines that term. A DataContext is lightweight and is not expensive to create. A typical LINQ to SQL application creates DataContext instances at method scope or as a member of short-lived classes that represent a logical set of related database operations.

Answer 3

static global DataContext? If my understanding of your question is correct, this will result in everyone connecting to you app sharing the same data context which will cause lot of security/sync issues. Avoid it.

Answer 4

DataContext discussion. Note I'm not commenting on your code.

DataContexts implement IDisposable, and therefore you should be disposing of the data context when it's no longer needed. Your website works well enough in development, but in production you will get nailed. You might as well do it right before your code gets too entrenched and changing it will be a big hassle. At best you'll just develop bad habits.

A better alternative to what you've written is to have your own controller base class that manages the lifetime for you.

public class MyBaseController : System.Web.Mvc.Controller
{
    private AbcDataContext abcDataContext;

    protected AbcDataContext DataContext
    {
        get 
        {   // lazy-create of DataContext
            if (abcDataContext == null)
                abcDataContext = new AbcDataContext(ConnectionString);

            return abcDataContext;
        }
    }

    protected override void Dispose(bool disposing)
    {
        base.Dispose(disposing);
        if (disposing)
        {
            if( abcDataContext != null )
                abcDataContext.Dispose();
        }
    }
}

which allows you to do

public class MyController : MyBaseController
{
    [AcceptVerbs(HttpVerbs.Post)]
    public ActionResult Edit(Item item)
    {
        DataContext.Items.Attach(item);
        DataContext.SubmitChanges();

        return View(item);
    }
}

While this works, I personally find it can get annoying and tricky.

---

Best: If you're going to follow MVC as you're supposed to, you should populate the model completely and not rely on lazy-loading entities. The best way to achieve this is to get rid of your DataContext as soon as you can.

Typically, we enforce this at a code level via the following pattern:

using( var dc = new AbcDataContext(ConnectionString))
{
    var itemUpdater = new ItemUpdater(dc);
    item = itemUpdater.Update(item);
}
return View(item);

The idea is that you will get an ObjectDisposedException if your view attempts to get any additional data via lazy-loading.

Answer 5

According to this discussion about the same problem, it seems to be a type-mapping bug that may be worked around if you delete the Item class in the designer and just drag over the table into the designer again.