To create a custom build activity you need to inherit from either CodeActivity or CodeActivity<T>, both of which reside in System.Activities. CodeActivity<T> is used to return a value out of the activity where T is the return type.

To get data into your class you need to create your properties using InArgument<T> (InOutArgument<T> and OutArgument<T> are also available). For example, if you need to get a file name into your class you would create a property: public InArgument<string> Filename { get; set; } which will appear as a string property named Filename in the workflow designer.

Use the attribute [RequiredArgument] to specify that a value is required for a property.

Finally you need to override the Execute method, this is the method that will contain the execution code for your build activity. The code activity context will be passed into this method which will allow you to access your properties. Once you have dragged the activity onto your build workflow you simply need to set the Workspace property to Workspace and set the FilePath property to the server path for the file you want to check out.

public sealed class CheckOutFile : CodeActivity
{
  [RequiredArgument]
  public InArgument<Workspace> Workspace { get; set; }

  [RequiredArgument]
  public InArgument<string> FilePath { get; set; }

  protected override void Execute(CodeActivityContext context)
  {
    // Obtain the runtime values
    Workspace workspace = context.GetValue(Workspace);
    String filePath = context.GetValue(FilePath);

    // Get the pending changes to the filePath
    PendingChange[] pendingChanges =
    workspace.GetPendingChanges(filePath, RecursionType.Full);

    // If no pending change for the filePath was found
    // it means the file is can be checked out
    if (pendingChanges.Length == 0)
    {
      // Check out the file
      workspace.PendEdit(filePath, RecursionType.Full);
    }
  }
}

Error #2048: Security sandbox violation: http://localhost/myapplication/main.swf cannot load data from http://remotemachine/webservice.

Not the most useful error message. Searching the web for this error produced a lot of useless results, eventually I discovered that I needed a crossdomain.xml policy file on the remote machine (in the web server root, in this case c:\inetpub\wwwroot) that grants the calling machine access. Where the crossdomain.xml file is in the format shown below (the domain attribute would obviously be replaced with one of the items mentioned in quotes).

<cross-domain-policy>
    <allow-access-from domain="URL" />
    <allow-access-from domain="URL Wildcard" />
    <allow-access-from domain="Specific IP Address" />
</cross-domain-policy>

My web service doesn’t know the URL or IP address of the application that will be calling it so I need to use domain=”*” to allow anyone to connect but if I put this in the cross domain policy file that is in the root it applies that access to everything below it not just my web service which is sitting in /webservice. This is obviously undesirable, so the solution seems simple – place the crossdomain.xml file in the /webservice directory.

In the code you would make the following call before making the request with the URLLoader class:

Security.loadPolicyFile("http://remotemachine/webservice/crossdomain.xml");

This tells the Flash framework to load a custom cross-domain policy file, simple right? No.
Unfortunately this did nothing, using Fiddler2 I could see that the custom cross-domain policy file was being requested at downloaded successfully, but wasn’t working for some reason.

Again, after searching through many useless results on the web claiming to describe how to fix the problem I discovered what to do, and it’s pretty simple (the links that I found are at the end of this article).

You need to create a cross-domain policy file (i.e. crossdomain.xml) in the root of the web server that allows custom cross-domain policy files and then Security.loadPolicyFile() will actually use the custom cross-domain policy file specified in /webservice.

You have a number of options for restricting how custom cross-domain policy files are used, the easiest and least restrictive is to allow all custom cross-domain policy files, to do this you would used the following:

<cross-domain-policy>
	<site-control permitted-cross-domain-policies="all" />
</cross-domain-policy>

The option I went with is to only allow policy files with a specific content-type, the master policy file (in the root of your web server) would contain the following:

<cross-domain-policy>
	<site-control permitted-cross-domain-policies="by-content-type" />
</cross-domain-policy>

Then I set up (in IIS) the extension .policy to have a MIME type of text/x-cross-domain-policy in the directory /webservice on my remote web server and created a file in the /webservice directory called crossdomain.policy with the following XML:

<cross-domain-policy>
    <allow-access-from domain="*" />
</cross-domain-policy>

If this all seems confusing there is the summary of what you need to do:

• Create a master policy file in the root of the web server allowing custom policy files.
• Create a custom policy file in the directory you want the policy to apply to.
• Call Security.loadPolicyFile(YOUR CUSTOM POLICY FILE) before calling urlLoader.load()

Why Adobe’s ActionScript Language Reference documentation for the Security.loadPolicyFile() method does not mention that you need to allow custom policy files in the master policy file is beyond me.

Links:
Cross-domain policy file specification
Policy file changes in Flash Player 9 & 10 – Workflows
Policy file changes in Flash Player 9 & 10 – Meta-Policies
A post on actionscript.org I created regarding my issue

Note: The solution should apply to Flash applications as well, not just Flex Applications, as all the classes mentioned above are actually Flash classes and not Flex classes. (I’m making an assumption here as I’ve never used Flash CSx, but I’m pretty certain it should )

You just need to add a block comment before your function that starts with two stars, i.e. /** */. Inside this block you can enter your comments about the function, if the function has parameters just include @param then the name of parameter and then the comment about it, and for return values you insert @return and the comment about the return value. See the example below:

public final class Strings
{
    /**
    * Checks if two strings are equal 
    * @param value1 The first string
    * @param value2 The second string
    * @param caseSensitive Should a case sensitive comparision be done, default is <b>false</b>
    * @return Returns true if the strings are equal otherwise returns false 
    */        
    public static function AreEqual(value1:String, value2:String, caseSensitive:Boolean=false):Boolean
    {
        // Function body here
    }
}

Doing this will display the following information when you navigate through the Intellisense list to the function you added the comment to:

Note you can include HTML tags in your intellisense comments, I’m not sure which tags are supported as I haven’t checked, I assume just the basic font formatting tags.

Also this works on fields, property getters/setters, etc.

Post based on Adobe Flash Builder 4 Beta 2 (I assume it is the same in Adobe Flex Builder 3)

In the launcher HTML template file you will see the following javascript:

var swfVersionStr = "${version_major}.${version_minor}.${version_revision}";
var xiSwfUrlStr = "${expressInstallSwf}";
var flashvars = {};
var params = {};
params.quality = "high";
params.bgcolor = "${bgcolor}";
params.allowscriptaccess = "sameDomain";
params.allowfullscreen = "true";
var attributes = {};
attributes.id = "${application}";
attributes.name = "${application}";
attributes.align = "middle";
swfobject.embedSWF(
	"${swf}.swf", "flashContent", 
	"${width}", "${height}", 
	swfVersionStr, xiSwfUrlStr, 
	flashvars, params, attributes);
swfobject.createCSS("#flashContent", "display:block;text-align:left;");

As you can see the flashvars and params are now passed into the Flash object as associative arrays, and the method that is called on the Flash object is no longer AC_FL_RunContent().

To get the query string into the flashvars object is very easy, simply iterate through the query string and add each key/value to the flashvars object.

var swfVersionStr = "${version_major}.${version_minor}.${version_revision}";
var xiSwfUrlStr = "${expressInstallSwf}";
var flashvars = {};

var queryString = window.location.search.substring(1);
if (queryString != null && queryString.length > 0)
{			
	var keyValuePairs = queryString.split("&");
	for (var i = 0; i < keyValuePairs.length; i++) 
	{
		var keyValuePair = keyValuePairs[i].split("=");
		
		if (keyValuePair.length == 2)
		{
			flashvars[keyValuePair[0]] = keyValuePair[1];
		}
	}
} 

var params = {};
params.quality = "high";
params.bgcolor = "${bgcolor}";
params.allowscriptaccess = "sameDomain";
params.allowfullscreen = "true";

var attributes = {};
attributes.id = "${application}";
attributes.name = "${application}";
attributes.align = "middle";

swfobject.embedSWF(
	"${swf}.swf", "flashContent", 
	"${width}", "${height}", 
	swfVersionStr, xiSwfUrlStr, 
	flashvars, params, attributes);
swfobject.createCSS("#flashContent", "display:block;text-align:left;");

Then to access the parameters inside your Flex application you just need to access the parameters property on the application object.

var someVariable:String = this.parameters["MyQueryStringKey"];
// OR
var someOtherVariable:String = FlexGlobals.topLevelApplication.parameters["MyQueryStringKey"];

That’s it, simple

Post based on Adobe Flash Builder 4 Beta 2

This post is basically just an introduction to what Flex and ActionScript are, and some random things I’ve learnt about them.

I have to admit when I was first told that I’d be working on a Flex application I wasn’t particularly happy, I’m not a big fan of Flash or scripting languages; but I have to say after working with it for a while I’m pleasantly surprised – it’s no .NET but for a scripting language it’s pretty good.

Flex is basically a library that sits on top of Flash providing developers an easy way to create Flash applications. For developers like myself the idea of having to create a Flash application with Flash CS4 is absolute hell; frames, movies, timelines and all that nonsense. So Adobe created the Flex library and the Flex IDE for developers like me that want to code in the traditional sense
I think Adobe have done a pretty good job given the constraints they were under building on top of the Flash framework; it’s not perfect by any means and it’s lacks a lot but much better than I was expecting.

Now I must make a clear distinction, ActionScript is Adobe’s implementation of the ECMAScript specification, so any limitations I mention are limitations of ActionScript and the ECMAScript specification; I’m assuming of course that Adobe has implemented the specification fully . Also, ActionScript is not a Flex thing, it is part of Flash.

Some random things about ActionScript 3.0:

• Requires Flash Player 9.0 or higher.
• There are no abstract classes or abstract methods.
• Does not have a concept similar to .NET generics or C++ templates
• Classes and class members can be private, protected, internal, or public.
• Has static class members but not static classes.
• No enumerated types.
• Properties are defined using get/set functions.
• Has interfaces, yay!
• Has exceptions and exception handling using try…catch…finally blocks.
• Anonymous functions are allowed.
• Has compile time type checking (it can be turned off if you really want it off )
• If you name your function/field the name as a class (case sensitive) you will need to fully qualify the type declaration.

Some random things about Flex:

• Collections/lists are weakly typed, i.e. Array and ArrayCollection hold the type Object.
• Has packages; which are a similar idea to .NET namespaces
• Internal does not have the same meaning as in .NET, internal in Flex means within the current package, not the current project.
• All file loading is done asynchronously with callbacks.
• The “event model” is weird (i.e. not true events), to “raise” events your class needs to inherit from EventDispatcher and classes hook into events on your class using the addEventListener method and specifying the string name of the event to hook into.
• You cannot define a signature for event handlers.

Just a quick disclaimer, I’m still new to Flex/ActionScript so hopefully everything I’ve mentioned here is correct

If you think I’m wrong about something, please let me know.

Before I continue let me just make it clear that I’m a .NET developer (I was a Delphi developer many years ago) and a fan of the Microsoft development environment so what I have to say is influenced by my experiences with Microsoft tools and languages.

While I think it’s admirable that Google wants to simplify programming, I personally don’t see anything enticing about the language to make the move to it. These are my problems with the language in it’s current form, who knows maybe future versions will be better

• No generics – I understand that it must be difficult to include generics in a compiler but come on, generics are so useful! (link)
• No exceptions – This I don’t understand! Come on, there are meant to be some sharp guys working on this project – how can they not include exceptions?! (link)
• No type inheritance – Now this is the one point that makes me think this language isn’t much better than VBScript. (link)
• No overloading of method or operators – This isn’t a big deal but it would be nice. (link)

So those are my main dislikes about the language, but there are some other problems too. I thought I’d give it a try even though I wan’t keen on it so I went to the download page and you have to download the source and compile the compiler, that’s where I lost interest in Go! How hard is it to provide binaries for the most common platforms! I’m assuming that it is just a command line compiler and that there is no IDE, which is another negative.

There are two things that they have done that are cool; one is that they have build in some clever support for concurrency using CSP, and the other is that Go will be available in Chrome and as a plugin for other browsers. This could be where Go finds its niche, personally I’d like to see a language more powerful than JavaScript in the browser.

I must also mention that Go is an experimental programming language so I’m probably being a little harsh

Adding this functionality is fairly easy; first add a base class to your LINQ to SQL entities as described in my post Adding a Base Class (or Interface) to All LINQ to SQL Entities and add the virtual methods OnSaving and OnSaved as below.

public abstract class EntityBase
{
    internal virtual void OnSaving(ChangeAction changeAction) { }

    internal virtual void OnSaved() { }
}

Then for any LINQ to SQL entities that you wish to have hooked up to these “events” implement the partial class and override the methods, for example you could use the OnSaving “event” to set your audit data:

public partial class Customer
{
    internal override void OnSaving(ChangeAction changeAction)
    {
        base.OnSaving(changeAction);

        switch (changeAction)
        {
            case ChangeAction.Insert:
                InsertedDate = DateTime.Now;
                break;
            case ChangeAction.Update:
                UpdatedDate = DateTime.Now;
                break;
            default:
                break;
        }
    }
}

Create a class called ChangeEntity and add the properties ChangeAction and Entity to the class, this class will be used later so that we know what action is happening to each entity.

internal class ChangeEntity
{
    public ChangeAction ChangeAction { get; set; }

    public EntityBase Entity { get; set; }
}

Now what we need to do is override the SubmitChanges method on the data context, the parameterless SubmitChanges method cannot be overridden as it is not virtual so we override the method SubmitChanges(ConflictMode failureMode); which the parameterless method calls.

We will then create a list of all the entities that have been modified (i.e. inserted, updated, or deleted) using the ChangeSet object that is returned by the GetChangeSet() method – the returned object has a seperate list for Deletes, Inserts, and Updates which we will join together into a single list.

Then we build up a list of ChangeEntity objects which will contain a reference to each entity to be changed and an enum of type ChangeAction stating the action to take place for each entity; namely Insert, Update, or Delete.

Then we enumerate through the list of ChangeEntity and call the method OnSaving on each entity.

Then we call SubmitChanges on the base class, and finally we enumerate throught the list of ChangeEntity once again and call the method OnSaved on each entity.

Below is the required code, it is more verbose that I would usually write but the original code didn’t display nicely

public partial class DataClassesDataContext
{
    public override void SubmitChanges(ConflictMode failureMode)
    {
        // Get the entities that are to be inserted / updated / deleted
        ChangeSet changeSet = GetChangeSet();

        // Get a single list of all the entities in the change set
        IEnumerable<object> changeSetEntities = changeSet.Deletes;
        changeSetEntities = changeSetEntities.Union(changeSet.Inserts);
        changeSetEntities = changeSetEntities.Union(changeSet.Updates);

        // Get a single list of all the enitities that inherit from EntityBase
        IEnumerable<ChangeEntity> entities = 
             from entity in changeSetEntities.Cast<EntityBase>()
             select new ChangeEntity()
             {
                 ChangeAction = 
                      changeSet.Deletes.Contains(entity) ? ChangeAction.Delete
                    : changeSet.Inserts.Contains(entity) ? ChangeAction.Insert
                    : changeSet.Updates.Contains(entity) ? ChangeAction.Update
                    : ChangeAction.None,
                 Entity = entity as EntityBase
             };

        // "Raise" the OnSaving event for the entities 
        foreach (ChangeEntity entity in entities)
        {
            entity.Entity.OnSaving(entity.ChangeAction);
        }

        // Save the changes
        base.SubmitChanges(failureMode);

        // "Raise" the OnSaved event for the entities
        foreach (ChangeEntity entity in entities)
        {
            entity.Entity.OnSaved();
        }
    }
}

There is a far simpler way to ensure that all LINQ to SQL entities inherit from a base class or implement an interface. Unfortunately there isn’t a way to do it from the LINQ to SQL designer, you need to open up the DBML file in a text editor. On the Database node simply add the attribute EntityBase along with the name of the base class or interface.

<Database Name="Northwind" 
	Class="MyDataContext" 
	xmlns="http://schemas.microsoft.com/linqtosql/dbml/2007"
	EntityBase="MyEntityBase">
</Database>

If the base class or interface is not in the same namespace as your LINQ to SQL entities then you will need to specify the full qualified name of the base class or interface.

Google Street View Car in Midrand, Gauteng, South Africa

My dad managed to stop the driver and have a chat with him, apparently Cape Town and Pretoria are done, and there are 4 teams working the Johannesburg area. No mention when it’s going live but I guess the driver wouldn’t know that.

Google Maps for South Africa has gone live though; check it out at http://maps.google.co.za/

So why would you choose to use one over the other? Well I guess it really depends on what you need; from my short investigation I was able to spot the following differences:

• Passing a null string into int.Parse() will throw an ArgumentNullException. Passing a null string into System.Convert.ToInt32() will return 0 (i.e. the default for int).
• int.Parse() has an overload where you can pass in the NumberStyles flags enum to specify how the string should be handled. System.Convert.ToInt32() does not have this overload.
• System.Convert.ToInt32() has an overload where you can specify the number base of the string, i.e. base 2, 8, 10, or 16.

Available overloads on int:

public static int Parse(string s)
public static int Parse(string s, NumberStyles style)
public static int Parse(string s, IFormatProvider provider)
public static int Parse(string s, NumberStyles style, IFormatProvider provider)

Available overloads on System.Convert (for the input type string):

public static int ToInt32(string value)
public static int ToInt32(string value, IFormatProvider provider)
public static int ToInt32(string value, int fromBase)

Some other points for you:

• If you don’t provide the IFormatProvider both int.Parse() and Sytem.Convert.ToInt32() will retrieve the format off the CurrentCulture.
• If you do not specify the NumberStyles for int.Parse() it will simply use NumberStyles.Integer.
• System.Convert.ToInt32(string) and System.Convert.ToInt32(string, IFormatProvider) both call int.Parse() to do the conversion, where as System.Convert.ToInt32(string, int) calls ParseNumber.StringToInt(). This is a static internal class so you will need to use Red Gate’s .NET Reflector if you want to see it. Something interesting about this call is that it makes an internal call to a method in the CLR to do the conversion.

Looking at the corresponding methods for conversion on the other types it appears they seem to work in a similar way, with the exception that System.Convert.ToDouble() and System.Convert.ToDecimal() do not provide an overload to specify the number base.

• Channel 9
  Channel 9 is the MSDN video website for all things dev related.
• dotnet.org.za
  Not a create site and most of the blog just rehash posts from other blogs but once in a while it has something useful.
• Eric Lippert’s Fabulous Adventures In Coding
  This is a blog I love, this guy works on the .Net Framework and posts some awesome goodies. If you want to be a good dev you should understand the framework’s internals and this blog will help with that.
• InfoQ (go to the site to create a custom feed)
  A lot of the stuff on here is high level, you won’t find many coding examples etc. But it has some great posts and the range of articles is very broad.
• Microsoft StyleCop
  This is just so that I know when StyleCop is updated.
• Mike Taulty’s Blog
  This blog pretty much focuses on SilverLight and Mike Taulty has a number of webcasts on SilverLight if you need to get going with it.
• Microsoft Architecture Center
  Just as the name says, Architecture related articles, some interesting, some boring.
• Mythical Man Moth
  A blog I discovered by another South African, I haven’t gone through all his posts yet but there are some interesting ones.
• rohland.co.za
  Another South African blog I discovered just a few days ago, I’m still going through the posts.

Then some blogs by various developers, I’m not going to comment on them, I’m sure you probably know most of these blogs already:

• Rick Strahl’s Web Log
• Scott Gu’s Blog
• Scott Hanselman’s Computer Zen
• Steve Sanderson’s Blog

If you’re frustrated and need a good laugh plug into this feed
ICanHasCheezburger

To do this you simply add the attribute configSource=”ConnectionStrings.config” to the XML node you want to replace with a file, and then in the file you’ve referenced you simply place the XML that would normally have been in your web.config file. This will allow you do deploy the web.config to different environments without overriding environment specific settings.

Example, in the web.config:

<connectionStrings configSource="ConnectionStrings.config" />

Then in ConnectionStrings.config:

<connectionStrings>
    <add name="ConnectionString1" connectionString="server=localhost; Integrated Security=True; Database=Database1;" providerName="System.Data.SqlClient" />
    <add name="ConnectionString2" connectionString="server=localhost; Integrated Security=True; Database=Database2;" providerName="System.Data.SqlClient" />
</connectionStrings>

Easy as that

Required Software

• Cruise Control .NET
• Subversion Console Client (e.g. Slik SVN)
• Microsoft IIS
• Microsoft Visual Studio 2008

Setup

• Make sure all of the required software is installed.
• Create the folder _Projects and bind it to the Subversion repository and perform an update.
• Modify the PATH environment variable to include the path to the folder
  C:\Program Files\Microsoft Visual Studio 9.0\Common7\IDE
  • Control Panel
  • System
  • Advanced Tab
  • Environment Variables
• Check that the virtual directory http://localhost/ccnet exists, if not then create it and point it to the folder C:\Program Files\CruiseControl.NET\WebDashboard
• Check that the service Cruise Control .NET Service is installed and running.

Cruise Control .NET Configuration

• Open the file C:\Program Files\CruiseControl.NET\Server\ccnet.config in a text editor. This file contains the settings for the different projects that will be monitored, the basic structure is as follows:

  <project>
      <workingDirectory /> 
      <triggers />
      <sourcecontrol /> 
      <tasks /> 
  </project> 
  

• Project Node: The project node just requires an attribute name with the name of the project.

  <project name="YourProject"></project> 
  

• WorkingDirectory Node: The workingDirectory node is the path to the folder where the project resides.

  <workingDirectory>
      D:\_Projects\trunk\YourProject
  </workingDirectory>
  

• Triggers Node: The triggers node simply specifies the interval Cruise Control .NET will use to poll the Subversion repository for committed files.

  <triggers> 
      <intervalTrigger seconds="60" />
  </triggers>
  

• SourceControl Node: The sourceControl node has the settings related to getting the source from Subversion.
  • The executable node is the path to the Subversion Console Client.
  • The workingDirectory node is the path to the folder where code from Subversion will be saved.
  • The trunkUrl node is the path to the project source in Subversion.
  • The autoGetSource node can always be set to true.
  • The username and password nodes are the username and password to connect to Subversion with.

  <sourcecontrol type="svn"> 
      <executable> 
          C:\Program Files\SlikSvn\bin\svn.exe
      </executable>
      <workingDirectory>
          D:\_Projects\trunk\YourProject
      </workingDirectory>
      <trunkUrl>
  
  http://localhost:8080/svn/Repository/trunk/YourProject
  
      </trunkUrl>
      <autoGetSource>true</autoGetSource>
      <username>david.turvey</username>
      <password></password>
  </sourcecontrol> 
  

• Tasks Node: The tasks node has a node msbuild that tells Cruise Control .NET to use msbuild.exe to build the project/solution along with all the settings needed.
  • The executable node is the path to MSBuild.exe
  • The workingDirectory node is the path to the folder containing the project/solution.
  • The projectFile node is the filename of the project/solution file.
  • The buildArgs node contains the arguments to pass to MSBuild.
  • The timeout node contains a value in seconds for the timeout of the build.
  • The logger node is the path to ThoughtWorks.CruiseControl.MsBuild.dll

  <tasks>
    <msbuild>
      <executable>
          C:\WINDOWS\Microsoft.NET\Framework\v3.5\MSBuild.exe
      </executable> 
      <workingDirectory> 
          D:\_Projects\trunk\YourProject 
      </workingDirectory> 
      <projectFile>YourProject.csproj</projectFile>
      <buildArgs>
          /noconsolelogger /p:Configuration=Debug
      </buildArgs>
      <targets></targets>
      <timeout>300</timeout>
      <logger>
          C:\Program Files\CruiseControl.NET\webdashboard
          \bin\ThoughtWorks.CruiseControl.MsBuild.dll
      </logger>
    </msbuild>
  </tasks> 
  

If you browse to the Cruise Control .NET dashboard (http://localhost/ccnet) your project should be in the list, probably with a state of unknown, simply click Force Build on the right to check whether your build succeeds or fails. When someone checks in code a triggered build will occur.

When dropping some object from the database, e.g. a table or stored procedure, the following statement is often used:

IF EXISTS (
        SELECT * 
        FROM sysobjects 
        WHERE type = 'P' AND name = 'GetAlertGeographies' )
BEGIN
    DROP PROCEDURE [Members].[GetAlertGeographies]
END 
GO

Now this might appear ok but it is not; the select statement that is checking the sysobjects table is not taking the schema into account. So this statement could attempt to drop the stored procedure even though it doesn’t exist because a stored procedure with the same name exists in a different schema.

This is the correct statement to use to take schemas into account:

IF EXISTS ( 
        SELECT TOP 1 NULL 
        FROM information_schema.routines 
        WHERE specific_name = 'GetAlertGeographies' AND specific_schema = 'Members' )
BEGIN
    DROP PROCEDURE [Members].[GetAlertGeographies]
END 
GO

If you want to check for a table just change “routines” to “tables”.

So here’s how you would do it with a try… finally block:

bool isUnique = false;
SqlParameter[] parameters = new SqlParameter[1];
Utilities.Database database = new Utilities.Database();
SqlDataReader dataReader = null;

try
{
    parameters[0] = database.MakeInParam("@FieldName", SqlDbType.VarChar, 50, _name);
    dataReader = database.GetDataReader("Accounts.ExtraDetailsFieldExists", parameters);
    if (!dataReader.HasRows)
    {
        isUnique = true;
    }
}
finally
{
    if (dataReader != null)
    {
        dataReader.Close();
    }
    if (database != null)
    {
        database.Dispose();
    }
}

return isUnique;

And to do the same thing using the C# using block:

bool isUnique = false;

using (Utilities.Database database = new Utilities.Database())
{
    SqlParameter[] parameters = new SqlParameter[1];
    parameters[0] = database.MakeInParam("@FieldName", SqlDbType.VarChar, 50, _name);
	
    using (SqlDataReader dataReader = database.GetDataReader("Accounts.ExtraDetailsFieldExists", parameters))
    {
        if (!dataReader.HasRows)
        {
            isUnique = true;
        }
    }
}

return isUnique;

Isn’t that far more readable? Far less code and its very clear where the scope of each object ends. Now I hear you shouting what about closing the data reader! Luckily for us the base class of the SqlDataReader closes the data reader when Dispose is called, and the same applies to the SqlConnection class.