Codebender

not the last of his kind

Follow me on TwitterRSS Feeds

  • Home
  • About Codebender
  • LiquidSilver SharePoint Framework

A singleton base class to implement the singleton pattern in C#

Mar 2nd

Posted by denni in Programming

No comments

The singleton design pattern is one of the most controversial pattern. Some people even call it anti-pattern, evil, or stupid. I won’t go into the debate now, but I just want to share something for those who need to use singletons.

Although it is one of the easiest pattern to understand, it is also very easy to code it poorly, leading to unexpected side effects such as poor performance or being not thread-safe.

Jon Skeet has written a very good article describing different implementations of the singleton pattern in C#. He gives a nice example on how you can write a singleton code that is thread-safe and lazy-instantiating. The code is less than 20 lines and you can use it as a template for your singletons.

Since in some projects I have created several singleton objects, it started to get repetitive. Therefore I decided to write a base class based on this pattern as follows:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
using System;
using System.Globalization;
using System.Reflection;
 
public abstract class SingletonBase<t> where T : class
{
  /// <summary>
  /// A protected constructor which is accessible only to the sub classes.
  /// </summary>
  protected SingletonBase() { }
 
  /// <summary>
  /// Gets the singleton instance of this class.
  /// </summary>
  public static T Instance
  {
    get { return SingletonFactory.Instance; }
  }
 
  /// <summary>
  /// The singleton class factory to create the singleton instance.
  /// </summary>
  class SingletonFactory
  {
    // Explicit static constructor to tell C# compiler
    // not to mark type as beforefieldinit
    static SingletonFactory() { }
 
    // Prevent the compiler from generating a default constructor.
    SingletonFactory() { }
 
	internal static readonly T Instance = GetInstance();
 
    static T GetInstance()
    {
      var theType = typeof(T);
 
      T inst;
 
      try
      {
        inst = (T)theType
          .InvokeMember(theType.Name,
            BindingFlags.CreateInstance | BindingFlags.Instance
            | BindingFlags.NonPublic,
            null, null, null,
            CultureInfo.InvariantCulture);
      }
      catch (MissingMethodException ex)
      {
        throw new TypeLoadException(string.Format(
          CultureInfo.CurrentCulture,
          "The type '{0}' must have a private constructor to " +
          "be used in the Singleton pattern.", theType.FullName)
          , ex);
      }
 
      return inst;
    }
  }
}

For an example of usage, we will create a singleton class that generates a sequence number:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// Must seal the class so no sub-classes can be defined.
public sealed class SequenceGeneratorSingleton
  : SingletonBase<sequenceGeneratorSingleton>
{
  // Must have a private constructor so no instance can be created externally.
  SequenceGeneratorSingleton()
  {
    _number = 0;
  }
 
  private int _number;
 
  public int GetSequenceNumber()
  {
	return _number++;
  }
}

Then, a console application will call the sequence generator class:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
class Program
{
  static void Main(string[] args)
  {
    Console.WriteLine("Sequence: " + SequenceGeneratorSingleton.Instance
	  .GetSequenceNumber().ToString());  // Print "Sequence: 0"
 
    Console.WriteLine("Sequence: " + SequenceGeneratorSingleton.Instance
	  .GetSequenceNumber().ToString());  // Print "Sequence: 1"
 
    Console.WriteLine("Sequence: " + SequenceGeneratorSingleton.Instance
	  .GetSequenceNumber().ToString());  // Print "Sequence: 2"
  }
}

To recap, in order to create a singleton class using the singleton base class, you need to do the following:

  1. Define a sealed class which derives fromĀ SingletonBase<T>, where T is the class name you are defining. It ensures that you cannot create subclasses from this singleton class.
  2. Define a single parameterless private constructor inside the class. It ensures that no instances of this class can be created externally.
  3. Access the class’ singleton instance and public members by calling theĀ Instance property.

I have also included the code in LiquidSilver, which you can view or download from the repository.

c#, design pattern, liquidsilver, tips

LiquidSilver switches to Mercurial on CodePlex

Mar 2nd

Posted by denni in News

No comments

As you may have noticed, CodePlex now supports Mercurial as the source control. Mercurial is a free and distributed version control system. It’s currently gaining a high momentum that big projects like Mozilla, OpenSolaris, OpenOffice.org, Symbian OS, and many others are now using it. Aside from CodePlex, other project hosting sites like Google Code, bitbucket, and SourceForge also offer Mercurial support.

Because DVCS has many advantages over traditional centralized VCS such as Subversion or CVS, I also decided to switch to Mercurial for managing the source code of LiquidSilver.

Setting Up Mercurial on CodePlex

If you’re starting a new project, CodePlex will give you the option to use Mercurial as your version control system. For existing projects, unfortunately, you cannot switch to Mercurial by yourself. Instead, if you are a project owner, you can contact CodePlex using the contact form and tell your intention to switch to Mercurial. Ensure that you’re logged in when doing this and mention your project name, user name, and the email address associated to your user name.

For me, it took less than 24 hours before CodePlex responded to my request and LiquidSilver had been migrated to Mercurial. All the versions and labels were intact and I could clone the repository successfully. Thanks, CodePlex.

Using Mercurial

If you are just starting and not familiar with Mercurial or any DVCS in general, it can be confusing at first since the concept is different. Fortunately, Joel Spolsky has composed a wonderful tutorial site called Hg Init that can very quickly and greatly help you understand DVCS or Mercurial in particular. I highly recommend you to visit it. You can also learn Mercurial from its official learning site or the book.

Although you can use Mercurial entirely from the command line interface, you may find it easier or more comfortable to use TortoiseHg that provides a friendly and familiar (if you have used TortoiseCVS or TortoiseSVN before) interface for Windows users.

Visual Studio users can benefit from VisualHG, a Visual Studio extension that can give you real-time visual status updates on your files from within the Visual Studio Solution Explorer window. VisualHG depends on TortoiseHg to execute the Mercurial commands.

Now, have fun with Mercurial.

codeplex, liquidsilver, mercurial, tips

Refreshing the LINQ to SQL DataContext by Clearing the Cache

Feb 8th

Posted by denni in Programming

No comments

LINQ to SQL DataContext can sometimes be difficult to deal with. Especially when combined with the use of stored procedures. For example, there is a case where I want to delete multiple records with a stored procedure rather than using the LINQ to SQL API for performance reason. After executing the stored procedure, the DataContext is not aware of the changes and still thinks that the entities are still there because the DataContext caches them. Hence, when I try to add another record using the same key, the DataContext throws an exception with the following message:

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

Logically, I want to refresh the DataContext memory so it will recheck the database. The provided DataContext.Refresh() method requires the original entity set to be sent which can be a problem. I don’t want to query the DataContext for the entities for performance reason, hence the stored procedure is used in the first place. So I have to find another way to refresh the DataContext or clear the cache without relying on the original entity set.

There is no official way to do it, but I stumbled on this blog post telling that there is a DataContext method called ClearCache() that can be used for that purpose. Unfortunately that method has internal modifier so you cannot call it from your code. But with a bit of reflection magic you can borrow the hidden power as follows:

context.GetType().InvokeMember(
	"ClearCache",
	BindingFlags.Instance | BindingFlags.Public | BindingFlags.NonPublic,
	null, context, null);

After invoking the ClearCache() method, I can now successfully add new entities to the DataContext by reusing the keys from the deleted records.

c#, linq to sql, tips
«12345»...Last »
  • Recent Posts

    • InfoPath 2010 breaks SharePoint 2007 installation
    • Common coding issues when working with anonymous SharePoint sites
    • Office 2010 breaks stsadm -o copyappbincontent
    • Flowchart: which open source license should I use?
    • System.Net.WebException: An exception occurred during a WebClient request
  • Recent Comments

    • sunny on Common coding issues when working with anonymous SharePoint sites
    • denni on Change Your Fonts, Change Your Mood
    • Deb Ramirez on Change Your Fonts, Change Your Mood
    • Yuva on Flowchart: which open source license should I use?
    • Office 2010 breaks stsadm -o copyappbincontent on InfoPath 2010 breaks SharePoint 2007 installation
  • Tags

    adobe best practices c# codeplex css design design pattern dynamic language flash flowchart font forefront freehand internet explorer javascript linq linq to sql linux liquidsilver mercurial office 2010 rdesktop security sharepoint subsonic tips visual studio windows windows 2003
  • Archives

    • User Login






      • Lost your password?
    • Links

      • Denni on LinkedIn My LinkedIn Profile
      • Denniland My personal blog
      • Ragnarok Online Database A database for the Ragnarok Online MMORPG
Mystique theme by digitalnature | Powered by WordPress
RSS Feeds XHTML 1.1 Top