How do you handle configuration differences for environment (dev, QA, prod) using Visual Studio?

When executing a software application, there are always differences in configuration, depending on the environment. These configuration changes may include references to a database server, contact email addresses, or how exceptions are handled. I was recently doing some research on how Visual Studio 2010 handles this.

The question I would like to pose to folks is:

"What is your thoughts and/or model for building and deployment of the software bits? How do you use Visual Studio or other tools to build and deploy a "kit" to an environment?"

Nested web.config files and connectionStrings (entry has already been added)

When using the Entity Framework (EF) recently, I ran into a situation where I received the following error message:

The entry 'myConnectionstring' has already been added...web.config line: 47

After investigating I found out this was due to my use of web.config files in nested directories. Here is the situation...

I have a reusable EF DAL (data access layer) that is used by several web services on a single web site. Several of these web services are in sub-folders with their own web.config files. This allows me to have custom settings for each service as well as being able to easily xcopy an entire directory and know I have all the correct settings. For the EF I have a connectionStrings key in the config files. In addition to the nested web.config files, I have the top-level config which also has the EF connectionStrings key.

When I tried to access a web service I would get the error above because .NET was merging the web.config files and found duplicate connectionString keys. To fix this I just had to add the following element within the connectionStrings section, directly before my key:

<remove name="myConnectionString" />

To facilitate knowing I can easily xcopy any web service, it is good practice to add this to all the web.config files, even the top-level one.

XDocument / XElement - Save using a StringWriter as UTF-8 rather than UTF-16

One thing I find myself having to do on frequent projects is to save XML to a text file. I tend to use the .NET XDocument or XElement classes and use the Save() method. Once and a while I need to call Save() using a TextWriter. By default, when I do this the XML is written using a UTF-16 processing instruction as:


<?xml version="1.0" encoding="utf-16"?>

More often than not you want UTF-8; .NET XElement won't even read the UTF-16 contents back in using the Load() method (and things like IE won't be able to view it in the browser). To work around this problem I derive a new StringWriter and override the Encoding method:

public class StringWriterWithEncoding : StringWriter
{
   private Encoding mEncoding;
   public StringWriterWithEncoding(Encoding encoding)
   {
      mEncoding = encoding;
   }

   public override Encoding Encoding
   {
      get { return mEncoding; }
   }
}

Now just use the new class and the XElement.Save() method to write as UTF-8:

XElement xml = XElement.Parse("<root/>");
using (StringWriterWithEncoding writer = new StringWriterWithEncoding(Encoding.UTF8))
{
   xml.Save(writer);
   Console.WriteLine(writer.ToString());
}

The row value(s) updated or deleted either do not make the row unique or they alter multiple rows

I was working with a Microsoft SQL database today (not mine thank goodness) that had duplicate rows; the table had no primary key defined. When I went to go delete the duplicate row (or even change it), I got the error:

The row value(s) updated or deleted either do not make the row unique or they alter multiple rows(2 rows)

It was somewhat hard to figure out how to delete the duplicate row. In the end, I was able to delete it by doing:

SET ROWCOUNT 1
DELETE FROM myTable WHERE statmentToSelectTheDuplicateRow

Another reason why to always define primary keys so you don't even get into this situation.

Remote Desktop using NLA from XP to Server 2008 R2

I recently had to connect to my Server 2008 R2 computer (only allowing Network Level Authentication (NLA) for RD) from an older XP client running SP3. No luck at first. Turns out that to use NLA from XP SP3, I had to make two registry changes.

I found out how to do this on https://support.soundenterprises.net/index.php?_m=knowledgebase&_a=viewarticle&kbarticleid=221, but I also included that information below just in case the source link disappeared one day.

---

Enabling Network Level Authentication on Windows XP Service Pack 3 for access to Server 2008 via Remote Desktop

When connecting to a Windows 2008 Server using remote desktop from a Windows XP client running service pack 2 or earlier, you get the following error message:

"The remote computer requires Network Level Authentication, which your computer does not support."

To enable NLA in XP machines; first install XP SP3, then edit the registry settings on the XP client machine to allow NLA.

Next, configure XP for NLA as follows:

1. Click Start, click Run, type regedit, and then press ENTER.
2. In the navigation pane, locate and then click the following registry subkey: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa
3. In the details pane, right-click Security Packages, and then click Modify.
4. In the Value data box, type tspkg. Leave any data that is specific to other SSPs, and then click OK.
5. In the navigation pane, locate and then click the following registry subkey: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\SecurityProviders
6. In the details pane, right-click SecurityProviders, and then click Modify.
7. In the Value data box, type credssp.dll. Leave any data that is specific to other SSPs, and then click OK.
8. Exit Registry Editor.
9. Restart the computer.

Server 2008 R2 / ApplicationPoolIdentity / Databases

I recently set up a new website on a Windows Server 2008 R2 computer running IIS 7.5 and SQL Server 2008. The new site would not run...wouldn't even start (.NET security issus, database login error). After some painful investigation on the issues, I encapsulated the steps to correctly use IIS 7.5's new ApplicationPoolIdentity support and allow access to a SQL Server 2008 database.

Microsoft will continue to cramp down on security and the use of ApplicationPoolIdentity is the default settings for a new website (so I'm blogging this to help others and to remind myself).

To the best of my knowledge, here are the steps to get your site up and running on 2008 R2 (if you have any corrections or feedback, please comment):

1) Using IIS, select “Application Pools”.  Right-click the pool to modify and
   choose “Advanced Settings”.  In the “Process Model” section:
     a. Verify “Identify” is set to ApplicationPoolIdentity.
     b. Verify “Load User Profile” is set to true.

2) Using IIS, select the site to modify.  In the IIS section, double-click
   “Authentication”.  Next “Edit” the “Anonymous Authentication” name and verify
   “Application pool identity” is chosen.

3) Using SQL Studio Management Studio, select “Security”, then “Logins”.  Add
   “New Login” using:
     a. Set “Login name” to “IIS AppPool\yourIISSiteName”.
     b. Select “Windows authentication” (don’t worry that this doesn’t resemble a
        real account on the computer; click OK).

4) Optional for security flexibility on the site files:
     a. Using Windows Explorer, right-click the site directory and choose
        “Properties”.  Select the “Security” tab and choose “Add”.
     b. In the “Enter the object names select” box, enter
        “IIS AppPool\yourIISSiteName”.  Click OK.
     c. Before leaving the “Permissions” dialog, customize the permissions for the
        new account (read, write, execute, etc.).

Whew!

Simulated Annealing - Poker Solitaire and C#

I recently revisited the Simulated Annealing heuristic algorithm using C#. This algorithm is traditionally the choice for solving the Traveling Salesman problem, but I wanted to tackle another large combinatorial problem, Poker Solitaire.

For a description of the problem, visit Dr. Dobbs original article at:

http://www.ddj.com/184408203?pgno=16

How do you calculate a result with a solution set of 25 factorial (15,511,210,043,330,983,907,819,520)?

The Visual Studio 2008 solution (see below) provides a console application that repeatable executes a single annealing process, each time trying to optimize a 5x5 matrix of playing cards. The algorithm attempts to lay the cards out to optimize 12 poker hands (5 rows, 5 columns, 2 diagonals).

You can execute the console application using the syntax below, passing it the name of a text file that represents 25 cards for the 5x5 matrix.

C:> SimulatedSolitaire.Console.exe TestData\BestCards.txt

Download the Visual Studio 2008 solution by clicking here.

Is "simple" better?

Just thought I'd pass along a lesson that I continue to learn.

A couple of years ago I had the need to know my ever-changing home IP address. Granted there are some free tools like Dynamic DNS and others that do this, but I had some special requirements that the IP had to be placed on another server for special reasons.

When I wrote the code, I created a service that ran on my home server. I used a C# interface so that I could "one day" move other types of data from my home PC to a server. I also had to have Visual Studio create the service installer and manually install the service. I created my Visual Studio solution with nicely decoupoled projects (assemblies). I was a genius.

Recently I had to change some stuff in that piece of code. The more I thought about it (and tried to remember all the code parts), I realized had I just taken a pure TDD (test-driven development) approach I could have done it much simpler. I then sat down and wrote a simple console app main() with about 4 lines of code to call a web page with a magical query parameter that did everything I needed. A quick job creation in Windows Task Manager and things were now running nicely.

Much less code and more importantly, much easier to maintain and make sense of. I know we all hear often (and I preach it as well) that we are to develop "extensible" code, but the reality of it is that probably 90% of the code we write will never even need to be extended. Given that, a TDD attribute is warranted in many cases (if not all of them).

I'm still learning that "simple is better" a lot of the time.

Prevent XSS attacks

I found this link very helpful in understanding and providing a good set of rules for protecting against Cross Site Scripting (XSS) attacks.

http://www.owasp.org/index.php/XSS_%28Cross_Site_Scripting%29_Prevention_Cheat_Sheet

jQuery 1.3 Cheat Sheet

Looking for a handy cheat sheet of jQuery objects and methods? Here is a pretty good two page guide in PDF format.

jQuery 1.3 cheat sheet (2 pages, 8.5" x 11")

How to install MVC 1.0 on a web server

Eventually you are probably going to want to install MVC on a web server. Assuming you do not have Visual Studio on that server, the command to install the MVC installation package is (all on a single command line):

msiexec /i AspNetMVC1.msi /q /l*v mvc.log MVC_SERVER_INSTALL="YES"

You will need to execute this command in an elevated command prompt window.