# Tuesday, August 26, 2008

There's nothing quite like riding on a bus for 27 hours with a few dozen software developers.

That's exactly what I did last weekend on my way to and from DevLink in Murfreesboro, TN.

The DevLink bus was the result of much hard work by Amanda who did most - if not all - the organizing.

The bus began its odyssey Wednesday night in Grand Rapids, MI before proceeding to Lansing.  When the bus picked me up in Plymouth, MI at 9:15 PM, it had already been traveling for 3 hours.  From there, it rolled on to Toledo, Columbus and Cincinnati (its final southbound stop at 4AM) before proceeding to its final destination in Murfreesboro, Tennessee. 

The first half of the ride was great fun.  I met new people; I reconnected with old friends; and I met people face-to-face with whom I had only communicated electronically.

Event the two times the bus driver got lost couldn't dampen our spirits.  We were having too much fun.

After about 4AM, the trip began to drag.  Everyone was exhausted and the seats were too uncomfortable to allow more than a few minutes sleep at a time.  This was clearly a bus designed for cross-town trips – not cross-country trips.  DVDs provided some entertainment ("The Big Lebowski" and “Office Space” cracked me up) but you can only focus on movies for so long and the acoustics were less than ideal.

We arrived at our hotel worn and weary at 9:30AM Central time, over 13 hours after leaving Plymouth and 16 hours after the bus began its trip. 

24+ hours of sleep deprivation left me physically ill and I spent nearly all of Thursday in bed.

It was good that I did.  The conference began Friday morning and I awoke refreshed and ready to absorb and exchange ideas at what turned out to be an excellent conference.  (You can read more about it here.)

The ride home was an adventure.  Although the conference ended at 6PM, someone decided the bus shouldn't leave town until 9PM so we had to kill a few hours at a restaurant before heading out. 

Although we picked up at least one new traveler on the ride home, we lost a few more.  Some folks booked a flight to Chicago for a Monday business meeting and a couple people elected to rent a car and drive home rather than subject themselves to the length and discomfort of the bus ride.  I was tempted to join them but I stuck with my plan and boarded the bus at 9.

The mood was more subdued on the return trip. We still had some good conversations early in the ride but the environment lacked the energy sparked by seeing people for the first time in months.

About 2AM, I had just begun to drift to sleep when a sound like a jackhammer awakened me suddenly.  I was sitting just above the tire that blew flat traveling 65 miles an hour on I-71 just north of Louisville, KY.  We pulled off the highway in front of a Waffle House in Carrolton, KY and waited 3 hours for the tire to be fixed.  The repair would have taken far less time if the repair guy had not allowed a drunken Waffle House customer to play with the lug nut gun.  He sheared off the bolts on wheel, forcing the repair guy to return to his shop for more parts and weld on new bolts.

Waffle House provided some entertainment.  Corey attempted to start a relationship with one of the waitresses; and the manager tossed out a drunk who wouldn't stop ordering pork chops long after he was told the restaurant was out of pork chops.  Apparently the Carrolton Waffle House is a magnet for drunks at 2AM Saturday night.  Go figure.

We got back on the highway about 5AM so exhausted that even the uncomfortable seats wouldn't prevent a few hours dozing.

I thought things would go quicker after this because the drivers knew the route better, having just driven it 3 days earlier.  Alas, we became lost in northeast Ohio and drove nearly to Cleveland before turning west toward Toledo.

At 12:30 Eastern time - 14.5 hours after leaving the hotel and 17.5 hours after the end of the conference - we pulled into the parking lot in Plymouth.  I was never so happy to see my car and the 40 minute drive home seemed trivial. 

I was scheduled to read aloud at the 12:15 mass in my church and the last minute substitute is probably still angry with me for missing this, but we all arrived safely.

Was the bus ride a success?  Would I do it again?  Would I take a bus to a distant location with dozens of others like me?  I've decide that, if I can be promised a more comfortable seat, I will do it the next time it's offered.  Everything else was easily tolerable and I did enjoy the fellowship that came with such a long ride with like-minded people.  I even got a few job leads from the conversations I had on the two rides (I am between jobs for those who don't know).

If you go on such a trip, my only advice is

  1. Be prepared for anything
  2. If the Waffle House waitress in Carrolton, Ky tells you they are out of pork chops, shut up and order the hash browns. 

Note: Click here to see more photos from DevLink and the famous bus ride.

Tuesday, August 26, 2008 4:38:58 PM (GMT Daylight Time, UTC+01:00)

People attend conferences for many different reasons.  Some come for the content of the lectures; some come to meet and hear well-known speakers; some come to meet and network with others in the industry; some come to see old friends.

Me, I come for all those reasons.  At DevLink last week in Murfreesboro, TN, I experienced all those things and more. 

But I also experienced something new.  I had heard of Open Spaces in the past but had not experienced them.  At DevLink, Open Spaces were promoted heavily as a different way of exchanging ideas.  I was curious and gave it a try.

An Open Space event consists of developers sitting together roughly in a circle in a room and they exchange ideas with one other.  A topic is picked in advance by the group but the conversation is not limited to that topic.  If the conversation drifts from the assigned topic and the group remains engaged, this is perfectly all right.  The important thing is that ideas are exchanged and the group remains passionate about the conversation.

And I heard a great deal of passion at the DevLink Open Spaces that I attended.

During the event, I attended 3 Open Spaces sessions plus the planning session (where topics were picked) and the wrap-up session (where the group reviewed the open spaces of the previous 2 days).  In each session I attended, I heard bright people sharing great ideas.  Sometimes we argued and sometimes we were in violent agreement but I enjoyed it all. 
In a session on Service Oriented Architecture, I argued earnestly that, due to the costs of SOA, support from the top was necessary for SOA to succeed within any organization.  Most of the other loud persons in the group insisted that newer tools such as WCF had lowered the cost of SOA sufficiently that a strong grass roots effort could drive SOA in an organization.  By the end of the session, I think we had all learned something and moved a little toward understanding the others' side.

I did attend a few traditional sessions in which a speaker stands in front of a classroom and delivers a lecture to an audience that is mostly passive.  Richard Campbell and Carl Franklin were two of the speakers at this conference and I have long been a fan of their .Net Rocks podcast, so I made a point to attend a lecture by each of them.  Both were good sessions but they were easily topped by Joe Wirtley who gave an excellent talk on WPF.  It was excellent because it focused on building a business application, rather than the eye candy that clutters so many WPF presentations.

Overall the conference was a great success.  It drained me of energy but it fired me up at the same time.

And I haven't even told you about the 28 hours I spent riding a bus with a few dozen techno-geeks.  Or the flat tire that left us stranded in Carrolton, KY for 3 hours at 2AM.  But that’s another story.

Note: Click here to see more photos from DevLink

Tuesday, August 26, 2008 4:49:27 AM (GMT Daylight Time, UTC+01:00)
# Friday, August 22, 2008

Microsoft recently released the ASP.Net Model View Controller framework (MVC).  It is currently available as Preview 3 and can be downloaded at http://www.microsoft.com/downloads/details.aspx?FamilyId=92F2A8F0-9243-4697-8F9A-FCF6BC9F66AB&displaylang=en

A new MVC project contains a couple sample views and controllers so you can get an idea of the proper syntax to use. 

In this article, we will add a new model, view and controller to an existing MVC project. 

1.       Open Visual Studio 2008 and create a new MVC project.  For information on how to create a new MVC project, see http://www.davidgiard.com/2008/08/18/TheASPNetMVCSampleAppDemystified.aspx

2.       Open the Solution Explorer (View | Solution Explorer) and select the Models folder.

3.       Add a Model to the project

a.       Right-click the Models folder and select Add | Class.  The Add New Item dialog displays.
Figure 1
    Figure 1

                                                               i.      At the Name textbox, enter “Customer”.

                                                             ii.      Click the OK button to create a Customer class.

b.      The Customer class opens in the class editor.  This class will contain a few public properties that help describe a customer object.  Add the following code to the Customer class.

    public class Customer


        public int CustID { get; set; }

        public string FirstName { get; set; }

        public string LastName { get; set; }

        public string StreetAddress { get; set; }

        public string City { get; set; }

        public string State { get; set; }

        public string PostalCode { get; set; }


4.       Add a Controller to the project

a.       In the Solution Explorer, right-click the Controllers folder and select Add | New Item.  The Add New Item dialog displays.
Figure 2
    Figure 2

                                                               i.      Under Categories, select Visual C#\Web\MVC.

                                                             ii.      Under Templates, select MVC Controller Class.

                                                            iii.      In the Name textbox, enter “CustomerController”.

                                                           iv.      Click the Add button to create the CustomerController class and open it in the class editor.

b.      In the CustomerController class, we will create some actions.  Each action will instantiate one or more Model objects and display them in a view object. 

                                                               i.      Add the following statement at the top of the CustomerController class. 

using TestMVC.Models;

                                                             ii.      Add the following private methods to the CustomerController class.  For now, we will create customers out of thin air (as if it were that easy).  In a real application, we would probably call a web server or query a database to get customers.

        #region private methods

        private Customer GetCustomer1()


            var cust1 = new Customer


                CustID = 1,

                FirstName = "David",

                LastName = "Giard",

                StreetAddress = "123 Main",

                City = "Boringville",

                State = "MI",

                PostalCode = "48108"



            return cust1;



        private Customer GetCustomer2()


            var cust2 = new Customer


                CustID = 2,

                FirstName = "John",

                LastName = "Smith",

                StreetAddress = "321 Elm",

                City = "Nowhere",

                State = "OH",

                PostalCode = "41001"



            return cust2;



        private List<Customer> GetAllCustomers()


            Customer cust1 = GetCustomer1();

            Customer cust2 = GetCustomer2();

            List<Customer> allCusts = new List<Customer> {cust1, cust2};

            return allCusts;




                                                            iii.      Add some Action method to the CustomerController class.  We’ll start with a List action.  Paste the following code into CustomerController.cs.

        public ActionResult List()


            var allCustomers = GetAllCustomers();

            return View("List", allCustomers);


                                                           iv.      In the List, we get a list of customers (all 2 of them) and return a view.  Unlike the generated code, we explicitly specify which view to return (“List”) and we pass in some extra data (allCustomers) that the view will consume.

5.       Add a view to the project.

a.       In the Solution Explorer, right-click the Views folder and select Add | New Folder.  A new folder appears in the Solution Explorer.  Rename this folder to “Customer”.

b.      In the Solution Explorer, right-click the Customer folder and select Add | New Item.  The Add New Item dialog displays.

                                                               i.      Under Categories, select Visual C#\Web\MVC.

                                                             ii.      Under Templates, select  MVC View Content Page.

                                                            iii.      In the Name textbox, enter “List”.

                                                           iv.      The Select a Master Page dialog displays. 
Figure 3
   Figure 4

1.       Navigate to the Views\Shared folder and select Site.Master.

2.       Click the OK button to add this view content page to the project.

c.       Add visual elements to the View

                                                               i.      If it is not already open, open the List view by double-clicking List.aspx in the Solution Explorer.  Click the Source tab at the bottom of the editor.

                                                             ii.      Replace the code in List.aspx with the following

<%@ Page Title="" Language="C#" MasterPageFile="~/Views/Shared/Site.Master" AutoEventWireup="true" CodeBehind="List.aspx.cs" Inherits="TestMVC.Views.Customers.List" %>

<%@ Import Namespace="TestMVC.Views.Customers"%>

<%@ Import Namespace="TestMVC.Models"%>


<asp:Content ID="Content1" ContentPlaceHolderID="MainContent" runat="server">

    <h2><%= Html.Encode(ViewData["Message"]) %></h2>












        <% foreach (Customer cust in (List<Customer>)ViewData.Model)

           { %>


                <td><%=cust.CustID %></td>









        <% } %>



The above code displays a list of Customer model objects.  In a real-world example, we may choose to have the Customer model derive from a base class and only refer to the base class in the view.  This would increase the separation between our view and our model.

6.       Test the application

a.       Save and compile the solution (Build | Build Solution).   Correct any errors you find.

b.  Run the solution (Debug | Start Debugging).  Depending on the port number used by Casini, it should display in your browser with a URL such as

c.       Navigate to the List page by changing the URL to
(Replace the port number if necessary)

d.      You should see a list of 2 customers in your browser
Figure 4
    Figure 4

e.      Set a breakpoint in the List method of CustomerController.cs and refresh the page to step through the code as it executes.

In this article, we created a model, view and controller from scratch and displayed them.  You can download the code for this project here.

In the next article, we will use the ID of the URL to specify a single customer.

.Net | ASP.NET | MVC
Friday, August 22, 2008 2:02:34 PM (GMT Daylight Time, UTC+01:00)
# Thursday, August 21, 2008

I have been plagued recently by a recurring problem with Visual Studio 2005.

When I attempt to exit Visual Studio, I receive the error "Visual Studio cannot shut down because a modal dialog is open" and I am not able to exit.  The rub is that there is no modal dialog - close every visible window in VS, close all other apps, resize, move and minimize VS and I can find no dialogs, modal or otherwise.  Some bit inside VS is set incorrectly convincing itself that a dialog has not yet been closed.

The only solution was to open Task Manager and kill Visual Studio.

I have automatic updates turned on so, if this is a bug, I expected it would have been fixed by now. 

I found a number of posts and threads about this issue so it is not uncommon but nearly every post did not list a solution.

After some digging, I discovered a hotfix for this problem that is not included in the normal Windows updates. 

You can read the details of the problem from Microsoft here http://support.microsoft.com/kb/936971 and you can download the hotfix here

One important point.  Visual Studio must be closed before installing the hotfix so you may end up killing the process via Task Manager one last time.

Thursday, August 21, 2008 12:51:23 AM (GMT Daylight Time, UTC+01:00)
# Tuesday, August 19, 2008

Every year, I become a bit more accepting of my own ignorance. 

I decided a long time ago that I wanted a career in which I could continue to learn and to expand my knowledge.   Software development affords me that opportunity because it is such a large field and because it changes so rapidly. 

It is conceivable (though extremely unlikely) I could learn everything there is to know about software development and find myself completely caught up with learning for one day.  If that miracle were to occur however, I would go to bed that night and awaken to discover that things had been invented while I slept.  And I would once again be ignorant about some things.

But that day will never come.  There is an infinite amount of knowledge to be acquired and a finite amount of time in which to learn it.  The ratio of what I know to what I don't know is likely to remain small.

Yesterday, I had lunch with a guy who knows a great deal about Windows Presentation Framework (WPF).  Since I am extremely ignorant about WPF, it was a good opportunity to learn some new things.  But WPF contains so much that we could probably have lunch every day for months and I would still only scratch the surface of this framework. 

I want to know everything but I realize and accept that I cannot.  So what's the answer? 

The answer is in learning how to find the answers.  "I don't know" is an acceptable answer to any question, but "I can't do it" is not.  As new challenges and problems arise, we need to be able to figure them out - to find the answer.  Usually our experiences only get us so far.  We need help finding answers. 

Certainly the World Wide Web helps.  Many times, when faced with a problem, I've discovered an article or blog post written by a developer who faced and conquered a similar problem.  Search engines such as Google allow us to find these solutions more quickly. 

Books and magazines help as well.  They provide knowledge based on the experiences of the authors.  So do Classes and conferences.

I've found one of the best ways to improve my knowledge base is to build a network of smart people on whom I can call for tough questions.  I try to reciprocate as much as I can, but luckily software developers tend to be very generous with their ideas.

So the conclusion I've come to after all this introspection is that what we know is not nearly as important as what we are capable of finding out. 

And that's encouraging for a guy like me who will never know it all.

Tuesday, August 19, 2008 4:59:22 PM (GMT Daylight Time, UTC+01:00)
# Monday, August 18, 2008

Microsoft recently released the ASP.Net Model View Controller framework (ASP.Net MVC).  It is currently available as Community Technology Preview 3 and can be downloaded at http://www.microsoft.com/downloads/details.aspx?FamilyId=92F2A8F0-9243-4697-8F9A-FCF6BC9F66AB&displaylang=en

This article describes how to create an ASP.Net MVC application and the code that is auto-generated for you.

Creating a new ASP.Net MVC project

1.       Open Visual Studio 2008.  Create a new project: Select File | New Project.  The New Project dialog displays.

    Figure 1

a.       Under Project Type, select Visual Basic\Web or Visual C#\Web, depending on your language preference.

b.      Under Templates, select ASP.Net MVC Web Application.  This application was added when you installed the ASP.Net MVC preview.

c.       Provide an appropriate location and name for the project and solution.

d.      Click the OK button to create the project.

2.       One of the advantages of an ASP.Net MVC project is that the separation of most of the code from the user interface makes it easier to write unit tests.  Visual Studio encourages you to create unit tests for your new project by prompting you with the Create Unit Test Project dialog every time you create an MVC project.

    Figure 2

a.       If you wish, you can decline to create a Unit Test project or you can change the default project name.  Typically I do not change any defaults on this dialog.

b.      Click the OK button to create the Unit Test project.

The Folder structure of an MVC project

When you create a new MVC project, Visual Studio, generates a couple views and controllers.  If you understand how these work, you can use them to guide how you will create more views and controllers.

The solution contains two projects: an MVC project and a unit test project.

View the projects in Solution Explorer.  Select View | Solution Explorer.  The MVC project contains several folders.

    Figure 3

1.       The Content folder contains a stylesheet Site.css for this site.

2.       The Controllers folder is where you will store all your controller classes.  By default, this folder contains a single controller class - HomeController.cs.

3.       The Models folder is where you will store any model classes for your application.

4.       The Views folder contains a subfolder for each view in your application.  By default, there are two subfolders: Home and Shared. 

a.       The Shared subfolder contains a master page for the site because it is shared by multiple web pages.  Any other UI elements shared by the site belong in this folder.

b.      The Home folder contains two pages: About.aspx and Index.aspx. 

5.       As with most web applications, the root folder of this project contains a Global.asax file and a Web.Config file, which contain setup and configuration information for the application as a whole.

The Files and Folders of an MVC project

Open Global.asax and view the code.  Notice that the Application_Start method (which fires once, at the startup of the web application) contains a call to the RegisterRoutes method. The RegisterRoutes method tells the MVC framework how to interpret a URL. 

public static void RegisterRoutes(RouteCollection routes)





        "Default",                                              // Route name

        "{controller}/{action}/{id}",                           // URL with parameters

        new { controller = "Home", action = "Index", id = "" }  // Parameter defaults




The routes.MapRoute method accomplishes this.  In this case, a “Default” route collection is created that interprets a URL with syntax like “{controller}/{action}/{id}”. 

·         The first part of the URL specifies the controller to use.  MVC looks in the Controllers folder for a class that inherits from System.Web.Mvc.Controller with a name that matches the controller specified in the URL.

·         The second part of the URL specifies the action to take.  The action is the public method within this controller that will be called. 

·         The third part of the URL specifies an id to pass to the action method.  This can be used to further customize the action.  For example, we could use the id as a filter to dynamically look up a single row in a database.

The routes.MapRoute method also allows us to specify defaults if no controller or action or id is specified in the URL.  If any of these are omitted from the URL, MVC will use the defaults specified in the third parameter of routes.Maproute.  In this case the object new { controller = "Home", action = "Index", id = "" } tells MVC the following:

·         If no Controller is specified in the URL, assume the Home controller (i.e., look for a class named “Home.cs” in the Controllers folder).

·         If no Action is specified in the URL, assume the Index action (i.e., look for a public method “Index” in the Home.cs class).

·         If no ID is specified in the URL, assume a blank ID (i.e., any code looking for an ID will retrieve an empty string).

Open the view files: Home.aspx and About.aspx and notice that there is no code behind in either.  This is because ASP.Net MVC applications do not execute the page life cycle.  All the code for this application is in the controllers.  These view pages contain only visual elements.

Open the controller class: Homecontroller.cs.  As we mentioned before, this class derives from the System.Web.Mvc.Controller class and it contains two methods: Index and About.

using System;

using System.Collections.Generic;

using System.Linq;

using System.Web;

using System.Web.Mvc;


namespace TestMVC.Controllers


    public class HomeController : Controller


        public ActionResult Index()


            ViewData["Title"] = "Home Page";

            ViewData["Message"] = "Welcome to ASP.NET MVC!";


            return View();



        public ActionResult About()


            ViewData["Title"] = "About Page";


            return View();




“HomeController” is the name of the class to implement the Home controller.  This is typical of how MVC works – developers follow naming conventions in order tell the framework where to find the code to run.  In the case of controllers, we implement a controller by sub-classing the System.Web.Mvc.Controller class, naming this subclass “xxxController” (where xxx is the Controller name) and placing that subclass in the Controllers folder of our MVC project.  If we wanted to call a controller named “David”, we would create a System.Web.Mvc.Controller class named “DavidController” into the Controllers folder.  This process is known as “convention over configuration”, meaning that the framework knows where to find code based on the names we use.

Let’s look closely at the Index method.  Recall that the method in the controller is the Action that is specified in the URL.  So the Index method will be called if the Index action is specified.

ViewData is a dictionary collection that is a property of every Controller object.  We can add or update items in this collection by syntax such as
ViewData["Title"] = "Home Page";

By placing items in this collection, we make them available to the view when it is called.

The view (remember this is the UI that the user will see) is returned from this method.  The following line returns the default view.
return View();

We know it is the default view because the statement did not specify the name of the view.  The default view has the same name as the Action called.  In this case, we are returning the Index view.  Once again, MVC uses conventions to determine where to find the view.  All views associated with a given controller are stored in a subfolder named for that controller beneath the Views folder.  In this case, we are using the Home controller, so we look for views in the Views\Home folder of the project.  The view itself is a file with the same name as the view and with an extension of “.aspx” or “.ascx”.  In this case, we are looking for the default view (Index) of the Home controller.  MVC renders the page Views\Home\Index.aspx for this view.  Again, the developer uses naming conventions to tell the framework where to find items.

Open Index.aspx.  Notice it displays the message stored in the ViewData dictionary by the controller.
= Html.Encode(ViewData["Message"]) %>

However, it contains no other code, because all logic is handled by the controller.


Creating a new ASP.Net MVC project is as easy as creating any other Visual Studio project.  Learning the paradigm that the MVC framework uses can be a challenge; but the samples created automatically with a new project can ease that learning curve.

In the next article, we will add a new controller and view to a project.

.Net | ASP.NET | MVC
Monday, August 18, 2008 4:37:58 PM (GMT Daylight Time, UTC+01:00)
# Sunday, August 17, 2008

The Model-View-Controller (MVC) design pattern has existed for years.  (http://heim.ifi.uio.no/~trygver/themes/mvc/mvc-index.html)  ASP.Net developers have been implementing it for years either through their own custom code or via third-party frameworks such as Monorail (http://www.castleproject.org/monorail/index.html).

Recently, Microsoft released the ASP.Net MVC framework to give web developers the option of using this design pattern without the need for a lot of “plumbing” code or the use of a third-party framework.

The Model-View-Controller design pattern splits an application into three distinct parts called (you guessed it) “Models”, “Views” and “Controllers”. 

A Model represents the stateful data in an application.  These are often represented as objects such as “Customer” and “Employee” that represent an abstract business object.  For persistent data, the Model may save and retrieve data to and from a database.  Public properties of these objects (for example, “LastName” or “HireDate”) represent their state at any give time.  The model objects have no visual representation and no knowledge of how they will be displayed on screen. 

A View is the application’s user interface (UI).  In a web application, this is the web page that the user sees and clicks and interacts with directly.  The View can display data but it has no knowledge of where the data it displays comes from.

A Controller is the brains of your application.  It links the Model to the View.  It handles communication between the other two parts of the application.  It is smart enough to detect when a data needs to be retrieved (from the model) and refreshed (in the view).  It sends updated data from the view back to the model so that the model can persist it. 

Below is Trygve M. H. Reenskaug’s diagram of the relationship between these thee parts.

    Figure 1

This separation of the various concerns of the application encourages developers to create loosely-coupled components.

Much of the communication to the controller occurs by raising events in the view, which keeps the controller loosely coupled from the other parts.  However the real advantage of the MVC pattern is that, because they only communicate through the controller, the model and view are very loosely coupled.  This provides the following advantages to an MVC application.

  • Loosely-coupled applications have fewer dependencies, making it easier to switch the user interface or backend at a later time.
  • Applications can be tested more easily, because so little code is in the view.  We can test our code by writing unit tests against the controller.

In the next article, we will look at how to create a Microsoft ASP.Net MVC application.

.Net | ASP.NET | MVC
Sunday, August 17, 2008 2:03:21 PM (GMT Daylight Time, UTC+01:00)
# Saturday, August 16, 2008
# Friday, August 15, 2008

Microsoft Visual Studio Team System 2008 Database Edition (aka “Data Dude”) provides tools for managing and deploying SQL Server databases. 

In this article, we will discuss how to migrate data from one database to another.   Data Dude provides the Data Compare tool for this purpose.

In order to use the Data Compare tool, the following conditions must be true

1.       Data exists in a source table.  You want to migrate that data to a table of the same name in a different database.

2.       Both tables must have the same structure.

3.       Both tables must have a primary key to uniquely identify rows.

Follow the steps below to migrate data with the Data Compare tool.

1.       Launch Visual Studio 2008

2.       Select Data | Data Compare | New Data Compare.  The New Data Comparison dialog displays

  Figure 1

a.       When migrating data from a table in one database to another, the database you intend to update is known as the “Target Database”.  The other database is known as the “Source Database”.  In the New Data Comparison dialog, select the Source Database and Target Database connections.  If you have not created a Visual Studio connection to these databases in Visual Studio, you can click the New Connection button to create them.

b.      The New Data Comparison dialog contains checkboxes that allow you to specify which rows you want to see and compare.  I don’t usually change these (they are all checked by default) but it may speed up the process to clear the Identical Records checkbox.

c.       Click the Next button to advance to the second screen of the wizard.

    Figure 2
On this screen, you can choose which tables to compare.  Usually I am only interested in one or two tables, so I clear the rest of the checkboxes.  If I have a million rows in my customer table and I’m not interested in migrating any of those rows, I can save a lot of processing time by un-checking the customer table.

d.      Click the Finish button to display the Data Compare window.

3.       The Data Compare window consists of two panes: the Object List on top and the Record Details at the bottom.

    Figure 3

a.       The object list displays each table or view as a single row with columns summarizing the number of new, removed, changed and identical rows.  Rows are matched on their primary key.

b.      Click a row in the object list to display details in the record details pane.  Here you can click a tab to view the rows that are new, missing or changed.  Checking the checkbox next to a record flags it to the Data Compare tool, meaning you want to update the target database to match the same row in the source database.  This may result in an INSERT, UPDATE, or DELETE statement depending on the tab on which the record is listed. 

c.       For a record to be flagged for update, both the table and the record must be checked.

4.       After checking all the rows you wish to update, click the Write Updates toolbar button to commit your changes to the target database. 

5.       Alternatively, you can click the Export To Editor toolbar button to generate a SQL script that you can run in the SQL Server query editor.  This method requires an extra step but has the following advantages

a.       You can modify the script before running it.

b.      You can send the script to someone else to run.

c.       You can view the script to learn what Data Dude is doing.  It’s interesting to note that constraints on each table are dropped before copying data, then created after the data is copied.  This speeds up the process.  Also, note the use of transactions to prevent incomplete data copies.  Below is a sample script updating data in one table.
This script was created by Visual Studio on 8/15/2008 at 8:57 AM.
Run this script on dgiard.Test_QA.dbo to make it the same as dgiard.Test_Dev.dbo.
This script performs its actions in the following order:
1. Disable foreign-key constraints.
2. Perform DELETE commands.
3. Perform UPDATE commands.
4. Perform INSERT commands.
5. Re-enable foreign-key constraints.
Please back up your target database before running this script.

/*Pointer used for text / image updates. This might not be needed, but is declared here just in case*/
DECLARE @pv binary(16)
ALTER TABLE [dbo].[OrderDetails] DROP CONSTRAINT [FK_OrderDetails_Orders]
ALTER TABLE [dbo].[OrderDetails] DROP CONSTRAINT [FK_OrderDetails_Products]
ALTER TABLE [dbo].[Orders] DROP CONSTRAINT [FK_Orders_Customers]
DELETE FROM [dbo].[ProductTypes] WHERE [ProductTypeID]=N'5646953f-7b89-4862-bcf3-bf53450d28bb'
INSERT INTO [dbo].[ProductTypes] ([ProductTypeID], [ProductTypeName]) VALUES (N'7beb0d99-d034-41b9-bbf7-f9cdcdbedc30', N'Furniture')
INSERT INTO [dbo].[ProductTypes] ([ProductTypeID], [ProductTypeName]) VALUES (N'abc19a14-5968-4c5f-9f0f-4debc034cb90', N'Hardware')
INSERT INTO [dbo].[ProductTypes] ([ProductTypeID], [ProductTypeName]) VALUES (N'b9e446ed-eeb1-4334-b191-c70a55ef1a05', N'Books')
ALTER TABLE [dbo].[OrderDetails] ADD CONSTRAINT [FK_OrderDetails_Orders] FOREIGN KEY ([OrderID]) REFERENCES [dbo].[Orders] ([OrderID])
ALTER TABLE [dbo].[OrderDetails] ADD CONSTRAINT [FK_OrderDetails_Products] FOREIGN KEY ([ProductID]) REFERENCES [dbo].[Products] ([ProductID])
ALTER TABLE [dbo].[Orders] ADD CONSTRAINT [FK_Orders_Customers] FOREIGN KEY ([CustomerID]) REFERENCES [dbo].[Customers] ([CustID])

The Data Compare tool is a simple tool for accomplishing a useful task.  Since I discovered it, it has saved me a lot of time setting up new data environments.


.Net | SQL Server | VSTS
Friday, August 15, 2008 2:02:53 PM (GMT Daylight Time, UTC+01:00)
# Thursday, August 14, 2008

Writing Unit Tests is an essential step in developing robust, maintainable code.  Unit Tests increase quality and mitigate the risk of future code changes.  However, relatively few developers take the time to write unit tests for their stored procedures.  The primary reason for this is that few tools exist to test stored procedures.

Microsoft Visual Studio Team System 2008 Database Edition (aka “Data Dude”) provides tools to help developers write unit tests against SQL Server stored procedures.  The tool integrates with MSTest, wich is a testing framework many developers are already using for their other unit tests.

In order to write unit tests for your stored procedures, those stored procedures must be in a database project.  For information on how to create a database project from a SQL Server database see: http://www.davidgiard.com/2008/08/11/DataDudeTutorial1CreatingADatabaseProject.aspx

This document describes how to create a database unit test.

1.       Launch Visual Studio and open your Database Project.

2.       Open the Schema View.  Select View | Schema View.

3.       Right-click a stored procedure and select Create Unit Test from the context menu.  The Create Unit Tests dialog displays.

    Figure 1

4.       Check the checkboxes next to all the stored procedures for which you wish to create unit tests.  Select the .Net language (Visual Basic .Net or C#) in which you want the automatic code to be generated.  You won’t be modifying this code so it isn’t that important, but I tend to keep all my code in the same language, so you may as well choose your favorite language here.  Enter a meaningful name for the Unit Test Project and class.  I like to name my Unit Test projects the same as my database project, followed by “Tests” or “UnitTests”.  If this is a new Database Unit Test Project, the Database Unit Test Configuration dialog displays.

    Figure 2

5.       The Database Unit Test Configuration dialog allows you to specify what you want to occur when you run these unit tests.  The dialog is organized into the following sections.

a.       Database connections

                                                               i.      Execute unit tests using the following data connection
This is the database against which tests will run.  Typically I set this to my Development or QA database.

                                                             ii.      Use a secondary data connection to validate unit tests
You may specify a different database to validate the syntax of your unit tests and test that all the objects you refer to exist.  I can only think this might be good if you are writing tests while disconnected from your testing database, but I never set this option.

b.      Deployment

                                                               i.      Automatically deploy the database project before unit tests are run
To save manual steps, you may wish to check this box and deploy the database project to the database each time you run your unit tests.  This slows down the testing step so I do not select this option.  I prefer to deploy my changes once; then run my unit tests – sometimes several times.

c.       Database state

                                                               i.      Generate Test data before Unit tests are run
It is often useful to populate your database with some test data prior to your test run.  Use this button to do this.

6.       After creating your unit tests, you need to modify each one and specify what you are testing.  Open the Solution Explorer (View | Solution Explorer).

7.       Double-click the unit test class to open it in the unit test designer. 

    Figure 2

8.       The Unit Test Designer contains some controls and two panes as described below. 

a.       A class can contain multiple tests.  The first control is a dropdown that allows you to select which test you are designing.

b.      To the right of the Test Name dropdown is another dropdown that allows you to specify what part of the test you are writing.  You can choose between the test itself, the “Pre-test” (which runs before the test is executed) and the “Post-test” (which runs after the test has completed – successfully or unsuccessfully).

   Figure 3

c.       Further to the right are three buttons that allow you to add a new test or to delete or rename the currently active test.

d.      Below the controls is the test editor.  This is where you will write your test.  You will write your test in T-SQL and Data Dude provides some stub code to get you started.  Write SQL statements that call your stored procedure and return one or more results.

e.      Below the test editor is the Test Conditions pane.  It is here that you enter your assertions. 

    Figure 4

                                                               i.      You can test for a given result set having 0 rows, 1 or more rows, or an exact number of rows. 

                                                             ii.      You can also test if a specific column and row in a given result set evaluates to a given value. 

                                                            iii.      Click the “+” button to add new assertions.  Highlight an existing assertion row and edit the row or click the “x” button to remove the assertion. 

                                                           iv.      Use the properties window to modify properties of the assertion.  Many assertions are based on a given resultset.  When I first started writing unit tests, I found it difficult to determine which resultset was which.  Basically, any line in your SQL script that begins with the word “SELECT” creates a resultset.  Each resultset is numbered, beginning with 1, in the order it is created in your script.  I sometimes find it useful to copy the SQL code and paste it into SQL Management Studio query window and run it.  Each resultset then appears in a separate grid in the Results pane.  Looking at these grids allows me to more easily see a sample result set and in what order they are created.

f.        You run your database unit tests the same ways you run any MS Test unit test.  One way to run the tests is to open the Test List Editor (Test | Windows | Test List Window), check the tests you want to run, and click the Run Checked Test toolbar button.  Tests in which all assertions prove true are passed; all others are failed.

Using Data Dude, you can extend your unit tests to cover your database objects and, therefore, improve the overall quality and maintainability of your code.

.Net | SQL Server | VSTS
Thursday, August 14, 2008 2:56:47 PM (GMT Daylight Time, UTC+01:00)