Materializing and attaching object graph is causing lazy-load SQL query for the original object

[Andrew]

My goal is to be able to attach an instance of a persistent object to a context which did not derive from an existing context or a SQL query (using IDbConnection).  

For instance:

//-------------------------------
//C# Class

public class Widget
{
    public int Id { get; set; }
    public int Name { get; set; }
}

//-------------------------------
//SQL Table

TBWIDGETS
ID INT PRIMARY KEY IDENTITY
NAME VARCHAR(100)

//-------------------------------
//C# Example

using (DbContext context = new DbContext())
{
    Widget detached = new Widget();
    detached.Id = 1;
    detached.Name = "test";

    //Causes a SQL query in the format:
    //SELECT NAME FROM TBWIDGETS WHERE ID = 1
    Widget attached = context.AttachCopy(detached);
}

I have attempted several different approaches to solve this problem include:

1. The above approach; utilizing the AttachCopy<T>() method

2. Using the context.Translate functionality against a DataTable / DataTableReader; derived from:

http://docs.telerik.com/data-access/developers-guide/low-level-ado-api/translating-dbdatareader/dev-guide-adoapi-translate-dbdatareader-persistent-types

Note: When I used the straight SQL query example in the above line, the materialization did not cause an additional SQL query.

3. Using the serialization / deserialization as documented in:

http://www.telerik.com/help/openaccess-classic/openaccess-tasks-working-with-objects-serialize-deserialize.html

While a single SQL query to "rebuild" the persistent class from the database is not too concerning, performing this operation for a collection will cause significant overhead.

 

Note: If the object is initially attached to context, detached and that detached instance is attached to another instance of the same context; this does not cause an additional SQL query.  Although in my scenario, I can not rely on this approach.

 

For reference, I am using version 2014.2.711.1 of the Telerik Data Access library.

 

Thanks