Dinesh's Blog ::::::: Being Compiled :::::::: .NET

Wanna images with patterns

Wed, 07 Jun 2006 04:00:43 GMT

Found these two images while surfing in .net and sql blogs. These were published by Raymond Lewallen [MVP]. I found that they are very important though it is little bit difficult to read. But worthwhile to keep them with you if you really interesting in OOP with patterns.

CSM – First Batch - ASP.NET completed

Mon, 03 Apr 2006 01:14:13 GMT

My first batch at CSM, completed their ASP.NET course, hope they got something from me. I tried as much as possible to give them more things related to C# and ASP.NET, sharing my experience. Sometimes I felt that the samples I used to explain some scenario are too complex for them but most of them grabbed it without mush problem. I couldn’t cover all the topics I wanted regarding ASP.NET because of the time but fundamentals and important topics. I just wanted to give them a feedback form – just to see how I have helped them – but couldn’t. I might be able to get a feedback from them when they come to exam . Good to see that some students are coming to my Windows class too.

WMI is not smart enough to find my HD serial number

Fri, 20 Jan 2006 01:36:45 GMT

I was used to get the volume serial number instead HD serial number for registration purposes but changed the plan couple of months back because of the possibility of changing the volume serial numbers. Then I have been using WMI for access my HD serial number and it worked fine until yesterday. Yep, it didn’t show the serial number of my new HD.

While surfing for new stuff, found a very good post “Revenge of DriveInfo” that access all I need with VB.Net code that is the translated of APIs. Seems this can be used for future application but need to test it carefully.

Have a look and see. It might be helpful to you all.

Colleagues don’t use this method. It is obsolete.

Mon, 02 Jan 2006 01:05:26 GMT

Think about an instance that you need a method to be hidden (in other words, keep the method as a reference) not to be deleted. How do you do it? I used to add region pre-processor directives but sometimes it made the code little bit complex. The best option is ObsoleteAttribute that takes care about this situation (I wonder why didn’t I use it). We can simply tag the method as “Obsolete” that gives compiler-warning after. Not only that, the method can be set as “Not to use” that gives compiler-error.

[ObsoleteAttribute("This method should only be used in DEBUG mode.")]
public void MyMethod1()
{

Console.WriteLine("MyMethod1 is exeuted");

}

[ObsoleteAttribute("This method should not be used.", true)]
public void MyMethod2()
{

Console.WriteLine("MyMethod2 is exeuted");
}

The warning (what we have set with positional parameter) will be displayed at the compile time for method1 but the second positional parameter that is added to the Method2 forces to compiler error.

Did you find it really useful?

Common mistakes we might make

Tue, 29 Nov 2005 01:47:23 GMT

Not qualifying instance members

This is a very common mistake we make. Have a look on below code sample.

class Employee
{
private string firstName;
private string lastName;

   public Employee(string firstName, string lastName)
   {
      firstName = firstName;
      lastName = lastName;
   }

   public void PrintName()
   {
       Console.WriteLine("Name is {0} {1}", firstName, lastName);
   }
}

Though we assign the firstName and lastName via constructor, values will not be set since the instance members are hidden by the parameters. One way to overcome this is, name the class members with “_” like “_firstName”. But I think the better way is, qualifying the class instance members with “this” keyword. This is one of the ways in which the “this” reference is typically used.

public Employee(string firstName, string lastName)
{
this.firstName = firstName;
this.lastName = lastName;
}

Remove items from the collection via looping with “for”

The below code loads instances of tickets into ArrayList and loops with “for”. Note that it sets the “IsValid” member to “true” for all tickets and removes the second ticket (which id is 2) from the collection. This remove causes inaccessibility of 3rd ticket (which id is 3) and because of this, 3rd ticket’s “IsValid” member will not be set as “true”.

public class Ticket
{
public int Id { get { return _id; } set { _id = value; } }
private int _id;

public bool IsFullyPaid { get { return _isFullyPaid; } set { _isFullyPaid = value; } }
private bool _isFullyPaid;

public bool IsValid { get { return _isValid; } set { _isValid = value; } }
private bool _isValid = false;

   public Ticket(int id, bool isFullyPaid)
   {
      this._id = id;
      this._isFullyPaid = isFullyPaid;
      this._isValid = false;
   }
}

ArrayList tickets = new ArrayList();
tickets.Add(new Ticket(1, true));
tickets.Add(new Ticket(2, false));
tickets.Add(new Ticket(3, true));
tickets.Add(new Ticket(4, true));

for (int x = 0; x < tickets.Count; x++)
{
((Ticket)tickets[x]).IsValid = true;

if (!((Ticket)tickets[x]).IsFullyPaid)
tickets.Remove(((Ticket)tickets[x]));
}

foreach (Ticket ticket in tickets)
{
Console.WriteLine(ticket.Id + ": " + ticket.IsValid);
}

Remember this behavior when you use “for” for collections. This is not an odd behavior but I have made this mistake for some codes and you might too.

Letting “new” to be used for derived methods

Wed, 21 Sep 2005 15:07:31 GMT

While searching the culprit for unexpected result generated by one of façades, found a small mistake I have made in one of my classes. Yes, the culprit is a derived method that is not preceded with either new or override keywords. This caused to compile the method with new keyword and, of course, it will always execute the base method instead of derived method if base class is used for declaration. Though to share the problem as well as the solution I made because it might help you on your next design. See the sample code.

public class Employee
{
protected string _name;

public virtual string Name { get { return "Employee name is " + _name; } }

public Employee(string name)
{
_name = name;
}
}

public class Manager : Employee
{
public Manager(string name)
: base(name)
{
}

public string Name { get { return "Manager name is " + base._name; } }
}

The “Employee” is the base class and Manager is the derived class. The expected result for below code is “Manager name is Dinesh” but the will resulted as “Employee name is Dinesh”. This is because of the “Name” property of Manager class is compiled with new keyword.

Employee employee = new Manager("Dinesh");
Console.WriteLine(employee.Name);

Now see the new code. This will give us the expected result because the method is preceded with “override”.

public class Manager : Employee
{
public Manager(string name)
: base(name)
{
}

public override string Name { get { return "Manager name is " + base._name; } }
}
}

Employee employee = new Manager("Dinesh");
Console.WriteLine(employee.Name);

The .net compiler will treat methods that are not preceded with “new” or “override” as “new” methods. But it seems that the behavior of java compiler is little bit differnet. I think that Java compiler will treat this kind of methods as overriden. Am I right?

Passing Custom Collection to the web service?

Sun, 17 Jul 2005 02:17:18 GMT

Passing Custom Collection via Web Service is impossible? For me, it seems it is impossible. The Custom Collection is represented as an Array instead of a collection at the client side (consumer side). Though it is as an Array at the client side, the web service works fine. Have a look in my code in web service.

[WebMethod]
public EmployeeCollection UpdateEmployees (EmployeeCollection employeeCollection)
{
   foreach(Employee employee in employeeCollection)
   {
      employee.LastName = employee.LastName + "Updated";
   }

return employeeCollection;
}

Sample classes I made for testing
public class Employee
{
private int _id;
private string _lastName;

public int Id {get {return _id;} set {_id = value;}}
public string LastName {get {return _lastName;} set {_lastName = value;}}
}

public class EmployeeCollection : CollectionBase
{

   public void Add(Employee employee)
   {
      List.Add(employee);
   }

   public void Remove(Employee employee)
   {
      if (List.Contains(employee))
         List.Remove(employee);
   }

   public Employee this [int index]
   {
      get
      {
         if (index >= 0 && index < List.Count)
            return (Employee) List[index];
         else
            return null;
      }
   }
}

Once the web service is added to the consumer application, The “UpdateEmployees” method is shown as;

WebSeriveConsumer.WebServiceTest.Employee[] ServiceTest.UpdateEmployees(WebSeriveConsumer.WebServiceTest.Employee[] employeeCollection)

I made a search on the internet for seeing a solution for this but not found a proper solution. One of the suggestions found was; changing the web service’s reference.cs file in consumer application. Yes, I changed the file (Adding a collection class to the reference.cs file and change all “Employee[]” to “EmployeeCollection”) and then it worked fine. I am not sure whether the way I changed file is correct or not because there is only a suggestion but information. But the biggest drawback with this solution is; it is needed to change the file again and again if the web service is updated.

Another issue I’m having with web service is, even though the class (either Employee or EmployeeCollection) is declared at the client side (consumer side), it cannot be used for passing to the web service because it is consider as a different object at the web service side. This causes to convert the object that is created with consumer declaration to web service type object. It seems this can be worked around by applying “Operator Overloading”. Is it a good idea? Or the declaration of the classes that use in the webservice at the client side is worthless?

Having lot of confuses in my mind, I started finding a solution for all of these with XMLSerialization. It might give me a nice solution for all. Anyway it is being tested.

Hope you guys experienced this type of problems and solved with some proper manner. Would you like to share your solution with me?

Assignment Operators cannot be overloaded, but "+=". How?

Tue, 12 Jul 2005 01:40:46 GMT

If you have played with "Operator Overloading" techniques, definitely you know the flexibility it turns on. For example, assume that you have a business object class called "Employee" and collection class called "EmployeeCollection" that hold Employee objects. The implementing "+" operator in "Employee", let you to have a code like below.

Employee employee1 = new Employee();
Employee employee2 = new Employee();
EmployeeCollection employeeCollection = employee1 + employee2;

Think that how nice it is if we can write something like this:

employeeCollection += employee1;
employeeCollection += employee2;

But the problem is the operator "+=". This cannot be overloaded. So, what's workaround? Simply use the operator "+" for "+=" too. See the code below.

//this allows to have code like: EmpployeeCollection = Employee1 + Employee2
public static EmployeeCollection operator+(Employee employee1, Employee employee2)
{
EmployeeCollection employeeCollection = new EmployeeCollection();
employeeCollection.Add(employee1);
employeeCollection.Add(employee2);

return employeeCollection;
}

//this allows to have code like: EmpployeeCollection += employee1
public static EmployeeCollection operator+ (EmployeeCollection employeeCollection, Employee employee)
{
employeeCollection.Add(employee);
return employeeCollection;
}

Done! Now the code: employeeCollection += employee1 is possible. How nice?

Wanna call server side methods asynchronously at client side? Use Ajax

Thu, 07 Jul 2005 04:10:31 GMT

Yesterday, Merill, at the dot net group meeting, came up with cool stuff called “Ajax” that contains a set of libraries that serializes .net objects and makes them available at client side. Basically, this allows us to call server side methods from the client side both synchronously and asynchronously. As I was looking for this type of library very badly, started exploring “Ajax” with some of my ASP.Net applications.

I started with various code samples and all of them worked fine. Here is a one of tests I did. This code tracks the dropdown “onchange” event and call server side method to get the relevant value asynchronously. (The logic behind is, when user selects a Doctor to make an appointment, shows the next available appointment number).

//Codes at web page

private void Page_Load(object sender, System.EventArgs e)
{
DropDownListDoctors.Attributes.Add("onchange", "CallGetNextAppoinmentNumber()");
Ajax.Utility.RegisterTypeForAjax(typeof(TestPage1));

}

[Ajax.AjaxMethod()]
public short GetNextAppoinmentNumber(short doctorId)
{
short appoinymentNumber;

//check with database

return appoinymentNumber;
}

   function GetNextAppoinmentNumber_CallBack(res)
   {
      document.getElementById("LabelAppoinmentNumber").innerText = res.value;
   }
</script>

As the method is called asynchronously, user can fill up the rest without any disturbance and see the appointment number even without submitting.

IE hit me with another restriction

Wed, 29 Jun 2005 03:44:51 GMT

One of my applications begins fail at certain level, yes, it is ASP.net application and the reason of failure is IE’s inability of handling nested tables. In one of web pages, dynamic nested tables are being created, earlier the number of nested tables created were around 10 to 20 but with the new extension, it becomes 10 to{ooops, no limit}. Unfortunately, IE handles only 27 nested tables and that makes me to re-think my design. I just had a search on this to see a light at the end of the tunnel but couldn’t find a single solution or sign of up coming IE version with more nested-tables capability. But the sad thing is “Mozilla” handles 33 nested tables.

See the first HIT at post.

Overriding ToString() vs New ToString()

Mon, 27 Jun 2005 02:25:43 GMT

Do you override ToString() method in your business classes? I know, most of the time we do not bother but sometime having an new version or overridden ToString() makes our life easier. Few examples are;

Have a look on this class:

public class Item
{
private short _id;
private string _name;

   public Item(byte id, string name)
   {
      this._id = id;
      this._name = name;
   }

   public string ToString()
   {
      return this._name;
   }
}

When you need to output a name of the item from an instance of Item, simply you can write;

Console.WriteLine(item.ToString());

Another example is; bind a list to DropDownList.

ArrayList list = new ArrayList();
Item item = new Item(1, "Item1");
list.Add(item);
item = new Item(2, "Item2");
list.Add(item);

DropDownList1.DataSource = list;
DropDownList1.DataBind();

Do we get the result expected? No. What is the problem? As we all know, DropDownList should call ToString() method at binding and display "Item1" and "Item2" instead of "YourApplication.Item" and "YourApplication.Item". This is because at the binding, it will call base.ToString() (in this case, base is an object class) instead of Item.ToString(). When we impement ToString() without specifying "new" or "override", it will compile as "new" version and will not override base.ToString(). If we write a code like below;

public override string ToString()
{
return this._name;
}

will result as we wanted. So, this is the difference between "new" and "override". Basically, when a method of base class is overridden by derived class, it will call derived method even with the base class implementation with derived class.

MyBaseClass myBaseClass = new MyDerivedClass();
myBaseClass.Method();

Max pool size was reached: It happened again

Fri, 24 Jun 2005 01:31:54 GMT

Have you ever experienced the error of “Timeout expired. The timeout period elapsed prior to obtaining a connection from the pool. This may have occurred because all pooled connections were in use and max pool size was reached”. You might have. Last week one of my clients complained that they experienced an error with their application for couple of minutes and worked fine after. The error that occurred was the same as that above mentioned. What is the reason for this? This is because; system is trying to store a connection object in the pool but pool cannot hold any more connections because it has reached the maximum connection objects level. I just noted down some points for connection pool for future references.

A pool is created based on the connection string for the connection. For each unique connection, ADO.net will create a connection pool.
When a connection is created and opened, ADO.net finds the relevant pool for the connection string and issues a released-connection (that is already added and not currently used). If not (either no connections in the pool or no released-connections), will add it to the pool if it can hold and issue.
If no more objects can be held and no more released-connections to be issued, an exception is thrown after the timeout period expired.
If the connection is not closed after the process, issued-connection will not be added back to the pool and remained as orphaned (un-released)
The default size of the connection pool is 100. If the connection is not closed, size of the orphaned connections in the pool is increased and overflowed.
The connection will not automatically add back to the pool when dispose. The connection should be explicitly closed. This behavior is little bit difference with binding concept (Connection might be closed when data source like data reader is bound to data grid).

Friends, if you can add more points, please comment.

Underlining AccessKeys in ASP.NET

Mon, 06 Jun 2005 04:46:20 GMT

How many you guys use “AccessKey” property of the web controls. I know that most are not using. The reason is, some are still do not know about the property or some do not use it because of its impossibility of setting underline key. It is true that though we can set the “AccessKey” of the control, we can’t show it to user unless we set either tool-tip or text with it.

Button1.AccessKey = "C";
Button1.Text = "Click";
Button1.ToolTip = "Click Alt+C to access this button";
//Button1.Text = "Click Alt+C to access this button";

The workaround for above problem is, use Html Button instead of ASP.NET button. By making the button as server control, we can do the same we do with ASP.NET. In addition to that, we can underline the access key.

By double clicking the button at the design window, Server click procedure can be generated or else we can use onserverclick attribute to handle the click event.

Accessing parent window via Popup window

Sat, 04 Jun 2005 02:34:12 GMT

We had been trying to access the parent window's objects via a popup but ended up with failure. Yesterday I just wanted to test this again and, so, I created a small web application and tested. Haha, found the mistake we have made. We have been trying with "parent" object instead of "opener". Anyway, I am really happy with it because this had caused lot of problems and had led us to find workarounds. If interest, here is the code.

Javascript in the WebForm1.aspx (form id is TestForm1) that opens WebForm2.aspx.

window.open("WebForm2.aspx");

Javascript in the WebForm2.aspx that accesses the value of the TextBox1 of WebForm1.aspx (parent).

alert(opener.TestForm1.TextBox1.value);

So, do not use "parent" instead of "operner".

Accelerate your coding speed with VS IDE Macro

Sat, 28 May 2005 01:27:05 GMT

Have you ever used macro with your VS IDE? If not, it is high time to use. So, how do we accelerate our coding speed with macro? In ASP.NET applications, one of the common statements is connection instantiation. If the connection string is stored in the web.config file, statement is similar to;

SqlConnection sqlConnection = new SqlConnection(System.Configuration.ConfigurationSettings.AppSettings["ConnectionString"]);

We need to write this statement every time needs to make a connection. How if this can automatically be written by single mouse click? Yes, we can. Simply create a macro that writes the statement to the active class. That's how the acceleration works. You can open the Macro explorer by clicking Alt+F8 or Tools->Macros->Macro Explorer menu. With that explorer, you can have your own module with new macros or can write macros in existing modules. But the worst case of this is, Macro IDE only support VB.NET (I usually don't code with VB.NET). Anyway I use couple of macros because it really makes our life easy. Here is one of the macros I heavily use. This is called "CreateCollectionClass" that creates a strongly-typed collection class for given name. For an example, if you need to have class called "EmployeeCollection" that derives from CollectionBase and implement Add, Remove and this methods, simply you have to run the macro by giving name as Employee. Here is the code for the macro.

Sub CreateCollectionClass()

   Dim className As String
   className = InputBox("Enter the class name")
   If (className <> "") Then

      DTE.ItemOperations.AddNewItem("Local Project Items\Class", className & "Collection.cs")
      DTE.ActiveDocument.Selection.StartOfLine(vsStartOfLineOptions.vsStartOfLineOptionsFirstColumn)
      DTE.ActiveDocument.Selection.PageUp(False, 5)
      DTE.ActiveDocument.Selection.EndOfLine()
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "using System.Collections;"
      DTE.ActiveDocument.Selection.LineDown(False, 7)
      DTE.ActiveDocument.Selection.StartOfLine(vsStartOfLineOptions.vsStartOfLineOptionsFirstText)
      DTE.ActiveDocument.Selection.EndOfLine()
      DTE.ActiveDocument.Selection.Text = " : CollectionBase"
      DTE.ActiveDocument.Selection.PageDown(False, 5)
      DTE.ActiveDocument.Selection.LineUp(False, 3)
      DTE.ActiveDocument.Selection.NewLine(2)
      DTE.ActiveDocument.Selection.Text = "public void Add(" & className & " " & LCase(className) & ")"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "{"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "List.Add(" & LCase(className) & ");"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "}"
      DTE.ActiveDocument.Selection.NewLine(2)
      DTE.ActiveDocument.Selection.Text = "public void Remove(" & className & " " & LCase(className) & ")"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "{"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "if (List.Contains(" & LCase(className) & "))"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "List.Remove(" & LCase(className) & ");"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "}"
      DTE.ActiveDocument.Selection.NewLine(2)
      DTE.ActiveDocument.Selection.Text = "public " & className & " this [int index]"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "{"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "get"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "{"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "if (index >= 0 && index < List.Count)"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "return (" & className & ") List[index];"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "else"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "return null;"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "}"
      DTE.ActiveDocument.Selection.NewLine()
      DTE.ActiveDocument.Selection.Text = "}"

End If
End Sub

When you run (double click on the macro created) the macro by giving a name as "Employee", will create a class called EmployeeCollection like below.

using System;
using System.Collections;

namespace yournamesapce
{
   /// <summary>
   /// Summary description for EmployeeCollection.
   /// </summary>
   public class EmployeeCollection : CollectionBase
   {
      public EmployeeCollection()
      {
         //
         // TODO: Add constructor logic here
         //
      }

      public void Add(Employee employee)
      {
         List.Add(employee);
      }

      public void Remove(Employee employee)
      {
         if (List.Contains(employee))
            List.Remove(employee);
      }

      public Employee this [int index]
      {
         get
         {
            if (index >= 0 && index < List.Count)
               return (Employee) List[index];
            else
               return null;
         }
      }
   }
}

See how easy is this. Writing macro is not much difficult because you can record what you do and save it as macro. You can make changes to the save file as you want too.

Assign shortcut key to the macro

This is also verys simple. Open the keybord category in the Environment in the option dialog box (Tools->Options menu). Serach the desired macro and assign a short-cut key. Make sure the key is marked as "global" and is not used for anything.

So, don't wait, accelerate your coding speed.

Unnoticed menu Items in VS IDE: FullScreen

Thu, 26 May 2005 16:36:43 GMT

I wonder why I didn’t use the “FullScreen” mode in VS IDE. By pressing Shift+Alt+Enter keys, “FullScreen” mode can be activated and it really makes the code page clear and wide (of course Full Screen). The moral of the above code is that there are some menu items still remain un-noticed by me. Have you gone through all menu items in the VS IDE?

Code of the PerformenceTimer

Thu, 26 May 2005 00:53:15 GMT

Received couple of requests to share the code of PerformenceTimer. So, here it is. Earlier I used to use Environment.TickCount property for handling timing information but found this code that uses QueryPerformanceCounter API. The QueryPerformanceCounter function retrieves the current value of the high-resolution performance counter, if one exists. The disadvantages of this are you might get 0 if the installed hardware does not support for high resolution performance counter and might get wrong result in multiprocessor environment. See MSDN for more information. Here is the code.

public class PerformenceTimer
{
   private long _startTime;
   private long _endTime;
   private long _frequency;

[DllImport("Kernel32.dll")]
private static extern bool QueryPerformanceCounter(out long lpPerformanceCount);

[DllImport("Kernel32.dll")]
private static extern bool QueryPerformanceFrequency(out long lpFrequency);

   public PerformenceTimer()
   {
      _startTime = 0;
      _endTime = 0;

       if (QueryPerformanceFrequency(out _frequency) == false)
       {
           //highperformence counter is not supported
           throw new System.ComponentModel.Win32Exception();
       }
   }

   public void Start()
   {
       QueryPerformanceCounter(out _startTime);
   }

   public void Stop()
   {
       QueryPerformanceCounter(out _endTime);
   }

   public void Reset()
   {
       _startTime = 0;
       _endTime = 0;
   }

public double ElapsedSeconds {get {return (double)(_endTime - _startTime) / (double)_frequency;}}

}

Does string always create a new instance when it is changed?

Sun, 22 May 2005 01:53:35 GMT

My answer would be "yes" if someone ask me before but now my answer is "no". As we know, strings are immutable and will create new strings on the heap when it is changed. But this scenario does not always apply. Especially with literal string, if the values of two strings are same, heap has only one string object but referenced by two variables.

string myString1 = "dinesh priyankara";
string myString2 = "dinesh " + "priyankara";
Console.WriteLine("myString1 and myString2 have same reference: {0}", Object.ReferenceEquals(myString1, myString2));

Output
myString1 and myString2 have same reference: True

This is how literal strings are managed. Internally all literal strings are stored in a table called intern pool and check the existence when new literal is added. If the same is exist, new string will not be created on the heap but referenced to existing one. But this is not the case with non-literal.

string myString3 = "dinesh";
string myString4 = "priyankara";
string myString5 = myString3 + " " + myString4;
Console.WriteLine("myString1 and myString5 have same reference: {0}", Object.ReferenceEquals(myString1, myString5));

Output
myString1 and myString5 have same reference: False

Though both myString1 and myString5 has same value, new string is created on the heap for myString5. But we can force to have a same object reference by using string.Intern method. This method will return the reference if one exists with same value else will add to the pool and return the new reference.

//string myString5 = myString3 + " " + myString4;
string myString5 = string.Intern(myString3 + " " + myString4);
Console.WriteLine("myString1 and myString5 have same reference: {0}", Object.ReferenceEquals(myString1, myString5));

Output
myString1 and myString5 have same reference: True

Now both myString1 and myString5 are referenced to the same object in the heap. If interest, read MSDN for more info.

Concatenation with Response.Write

Fri, 20 May 2005 01:11:00 GMT

This is very simple thing but important. If we have couple of strings to be written out, what we were (I was) used to do is, concatenating strings and output with single Response.Write call. But this is not the way if we consider about performance and have heavy concatenation. I found this statement while reading some points regarding ASP.NET; "using multiple Response.Write is faster than concatenation with single call". The code below is the test I did myself.

//string out with single call

string string1 = "string 1 ";
string string2 = "string 2 ";
string string3 = "string 3 ";
string string4 = "string 4 ";
string string5 = "string 5 ";
string string6 = "string 6 ";
string string7 = "string 7 ";
string string8 = "string 8 ";
string string9 = "string 9 ";
string string10 = "string 10 ";

PerformenceTimer.PerformenceTimer timer = new PerformenceTimer.PerformenceTimer();

timer.Start();
Response.Write("concatenation "+string1+string2+string3+string4+string5+string6+string7+string8+string9+string10);
timer.Stop();

Response.Write("<br>");
Response.Write("Time elapsed: ");
Response.Write(timer.ElapsedSeconds.ToString("N", numberFormatInfo));

output
concatenation string 1 string 2 string 3 string 4 string 5 string 6 string 7 string 8 string 9 string 10
Time elapsed: 0.0000064254

ran the application several times and the time elapsed around .0000060000 - .0000070000.

//string out with multiple calls

PerformenceTimer.PerformenceTimer timer = new PerformenceTimer.PerformenceTimer();

timer.Start();

Response.Write("non concatenation ");
Response.Write(string1);
Response.Write(string2);
Response.Write(string3);
Response.Write(string4);
Response.Write(string5);
Response.Write(string6);
Response.Write(string7);
Response.Write(string8);
Response.Write(string9);
Response.Write(string10);

timer.Stop();
Response.Write("<br>");
Response.Write("Time elapsed: ");
Response.Write(timer.ElapsedSeconds.ToString("N", numberFormatInfo));

output
non concatenation string 1 string 2 string 3 string 4 string 5 string 6 string 7 string 8 string 9 string 10
Time elapsed: 0.0000044698

did the same times and the time elapsed around .0000040000 - .0000045000.

so, there is a big difference. We will not see the difference if few strings need to be handled but better add this way to best practices.

Accessing .NET assembly through DTS

Fri, 13 May 2005 01:15:13 GMT

This was terrible because it kept me awake whole night. Yesterday I wanted to access a function I have written with c#.net through SQL Server DTS. So, simply, registered the assembly in the register by using "regasm.exe" and tried to access it but failed. As I was at the customer site, couldn't do thorough search on the web but went through many sites and studied samples given at the night. Finally I found the problem: the "/codebase" argument is NOT used that specify the path of the assembly if it is not installed in the GAC. I know, it is a silly mistake but made it. So, thought to blog it.

This is my test class.

namespace DotNetClassLibrary
{
public class TestClass
{

   public TestClass()
   {
   }

   public void TestMethod(string fileName, string message)
   {
      StreamWriter streamWriter = File.AppendText(fileName);
      streamWriter.WriteLine(message + ": " + DateTime.Now.ToString());
      streamWriter.Flush();
      streamWriter.Close();
   }
}

This should be signed with strongly-name. So, created a "snk" file with "sn.exe" and placed to the project folder.

[assembly: AssemblyKeyFile("..\\key.snk")]

Copied the dll to the folder called "DotNetLibraries" and registered the assembly.

E:\ DotNetLibraries>regasm.exe DotNetClassLibrary.dll /tlb: DotNetClassLibrary.tlb /codebase

Done! Now this can be simply accessed by the ActiveX task in the DTS.

Function Main()
dim dotnetobject
set dotnetobject = createobject("DotNetClassLibrary.TestClass")
dotnetobject.TestMethod "E:\\Result\\Test1.txt", "this is test"
Main = DTSTaskExecResult_Success
End Function

But still one problem remains. Though it can be un-registered, seems it didn't completely remove. Again went through some blogs and found couple of good posts (This has something good). Tonight, I have to find the second mistake I have done.

The beauty of "readonly"

Thu, 12 May 2005 01:40:43 GMT

I had a class library that has couple of constant fields that was used in one of my freelance application. Recently, my client wanted to make some changes for the application and, yes it caused to change the class library. So, changed values of some const fields and uploaded the dll of the library. Guess what? New values of const fields are NOT applied. What was the problem? So, made some searches, and found something I was unaware (I should have known this). "All the const fields are replaced to its value when the application is compiled." Though I have built the library and replaced, as the main application is not compiled again, it still has the old value. So, then I rebuilt the application and fixed the problem. And I found another thing, if I had used "static readonly", the application is NOT needed to be recompiled. To test and make sure, I wrote a simple application. Here it is.

This is the class in the library with const field

namespace ClassWithConstant
{
   public class Class1
   {
     public const int constInteger = 150;
     public Class1() {}
   }
}

This is the class in another library with static readonly field

namespace ClassWithReadOnly
{
   public class Class1
   {
      public static readonly int readOnlyInteger = 250;
      public Class1(){}
   }
}

Code in my main application that has references for both above classes (dlls)

[STAThread]
static void Main(string[] args)
{
Console.WriteLine("Value of const of referenced class: {0}", ClassWithConstant.Class1.constInteger);
Console.WriteLine("Value of readonly of referenced class: {0}", ClassWithReadOnly.Class1.readOnlyInteger);
}

Then placed the release of the main application to the different location. Here is the output of release.

Value of const of referenced class: 150
Value of const of referenced class: 250

After that, changed the code of the both classes in the both libraries and just replaced the "dll" in the new location. Now, here is the new output.

public const int constInteger = 1500;
public static readonly int readOnlyInteger = 2500;

Value of const of referenced class: 150
Value of const of referenced class: 2500

See, still the const filed has old value but static readonly. You can get some good points from http://community.sgdotnet.org/blogs/triplez/archive/2005/04/14/13768.aspx.

IsNumeric with c#.net

Wed, 11 May 2005 01:00:26 GMT

Couple of months ago, I found a good "VB.NET IsNumeric" equivalent for c#.net while was searching an equivalent. I had an equivalent that implemented with "try" and "catch" blocks but decided to stop that way and stick into new way. So, thought to share it.

//my old way
public static bool IsNumeric(string text)
{
   try
   {
       int integerValue = Convert.ToInt32(text);
   }
   catch
   {
       return false;
   }
   return true;
}

//testing
string text = "sdfds dffds fa45 g68";
PerformenceTimer.PerformenceTimer timer = new PerformenceTimer.PerformenceTimer();

timer.Start();
bool isNumeric = Test.IsNumeric(text);
timer.Stop();
Console.WriteLine("Is numeric: {0}. Elapsed time: {1}", isNumeric, timer.ElapsedSeconds);

output
Is numeric: False. Elapsed time: 0.0049065

//method now using
public static bool IsNumeric2(string text)
{
   char[] charArray = text.ToCharArray();
   for(int x=0; x<charArray.Length; x++)
   {
    if (!char.IsDigit(charArray[x]))
        return false;
   }
   return true;
}

//testing
string text = "sdfds dffds fa45 g68";
PerformenceTimer.PerformenceTimer timer = new PerformenceTimer.PerformenceTimer();

timer.Start();
bool isNumeric = Test.IsNumeric(text);
timer.Stop();
Console.WriteLine("Is numeric: {0}. Elapsed time: {1}", isNumeric, timer.ElapsedSeconds);

output
Is numeric: False. Elapsed time: 0.0013188

The “ref” and “out” modifiers in method arguments

Fri, 06 May 2005 01:45:14 GMT

I was asked by one of my friends that what’s different by passing values as “ref” and “out” if both are passed “by reference”. At once I couldn’t answer but said that with “out” modifier, parameters can be passed un-initialized. Anyway, thought I’d better brush up my knowledge, did a small code.

//This method accepts two paras, one as ref and one as out
public void Method01 (ref int y, out int z)
{
y++;
z = 3;
}

int y = 2;
int z;

MyMethod(ref y, out z);

The z variable is not initialized but can be passed to the method. But a value must be assigned to z within the method. And y must be assigned before passing it to method.

SortedList: The "Hybrid" between a Hashtable and an Array

Wed, 04 May 2005 19:20:43 GMT

Though I was aware about SortedList, I have never ever used it (Or I was unaware of its benifts) for coding. But today it really helped me because I had a dictionary type collection that needed to be sorted by its key. Because the elements in the SortedList are autometically sorted by its key and values can be accessed through both index and key, it really cut down some my heavy work load.

Some key points about SortedList

Does not allow duplicate keys.
IComparer/IComparable can be implemented for sorting instead of default.
Might decrease performance as need to re-organize for every inserting and removing.

Sorting values by size in ArrayList

Wed, 04 May 2005 19:14:00 GMT

One of my colleagues came and asked me some best ways to sort elements (string type) by size in an ArrayList. Obviously, there are couple of ways to do the sort (he had some in his mind too) but needed to make it simple. So, I asked him to have a Comparer class that do the compare base on the size and use it for "Sort" method. And it worked……Here is the code.

public class LengthComparer : IComparer
{
   public int Compare(object x, object y)
   {
       return (new CaseInsensitiveComparer().Compare(y.ToString().Length, x.ToString().Length));
   }
}

ArrayList arrayList = new ArrayList();
arrayList.Add("lmn");
arrayList.Add("abcde");
arrayList.Add("wxyz");

arrayList.Sort(new LengthComparer());