Understanding .NET Reference Types and Value Types

I have been developing in .NET now for about 4 years and while I have visited this topic many times I have not always fully understood things. Dealing with Value Types is really straight forward however when we start to get into Reference Types things become a bit clouded. Well I have recently been forced to try and explain them to someone else and so I decided it was time to make sure I fully understand things. So this is my attempt to provide based on things I have read and learned how things work.

 I will start by simply making some statements that will later help us understand things.

Value Types - Stored on the stack. When you declare a variable based on a value type, the CLR allocates memory for the value type instance wherever the variables is declared.

Reference Types - Variable declared is stored on the stack. When you declare a reference type, the CLR allocates memory for instance on the heap.

Reference Types with Value Types: When declaring inside a reference type a field that is a value type it is important to note that the memory is allocated inside the hosted object.

-----------     ------------
-  stack  -     -  heap    -
-----------     ------------
-  var1   -     -  Class1  -
-----------     - Instance -
-   10    -     ------------
-----------           ^
-  var2   -           |
-----------           |
-   ref   - -----------
-----------

Strings - It is important to point out that strings are immutable. When making changes to a string you are actually creating a new instance of that string. You will see in the examples below that they behave a little different when making changes to them as opposed to making changes to a field in a Class.

Memory Cleanup:

  • Value Types - When a value type goes out of scope the memory is immediatly removed from the stack.
  • Reference Types - When a reference type goes out of scope the variable is immediatly removed from the stack. However the memory allocated on the heap still exists and the removal is managed by the Memory Manager and Garbage Collector process.

 Methods: Passing arguments to methods ByValue and ByReference When you declare a method you by default pass in the argument ByValue. However the behaviour giving this statement is not always as expected when dealing with reference types. Here are some scenarios.

  • Value Type being passed ByValue: Remember that a value type holds the data itself. When passing in ByValue a copy of the actual data is passed. The method now contains its own variable and data and so the result is that the original variable will not change as changes are made to the new variable inside the method.
  • Value Type being passed ByReference: When passing in a value type ByReference we are actually passing a reference is created that points to the data. The result is that any changes made to the variable inside the scope of the method are reflected back to the external variable.
  • Reference Type being passed ByValue: When you pass in a reference type ByValue you get a bit  different behaviour. Remember that the value is always on the heap, so you are actually pointing the new variable to the same memory location as the original variable. Thus when making changes to the variable changes are reflected to the original variable. So if you need to pass a reference type and not have changes reflected you need to make sure you create a full new copy of the object being passed in.
  • Reference Type being passed ByReference: This last situation you most likely want to avoid. This basically ends up creating a reference variable that points to the reference variable that points to the data. So a reference to a reference. While I said you usually should avoid this it is worth pointing out why you might want to use this approach. Remember that in the  previous Reference ByValue changes are reflected back to the original variable. However if you make any changes to the object itself, for example set the object to NULL or set it to a NEW object then the changes would not be reflected outside as your internal variable now points to a new memory location. So if you need to have NEW or NULL changes reflected back outside you could pass it in ByRef.

 Below you will find some examples that will help see how value types and refernce types differ when making changes to them and how the behaviour changes when passing them into methods.

STRING TEST - Reference Type - Immutable
Creating two strings. Setting the first to the value "one-before" and then initializing the second variable by pointing it to the first. Writing out the initial values so that we can see that each is now set to "one-before". So at this point what we have in effect is to variables declared on the stack one and two that both point to the same memory location which holds "one-before". Then we come along and we set one = "one-after" and since strings are immutable a new memory locations is allocated and one is now pointed to the new memory location. However two still remains pointed to the previous memory location so the result is that they now have different values.

//Begin Test
string one = "one-before";
string two = one;

Console.WriteLine(one); //Output = one-before
Console.WriteLine(two); //Output = one-before

one = "one-after";

Console.WriteLine(one); //Output = one-after
Console.WriteLine(two); //Output = one-before
// End Test


INTEGER TEST - Value Type
Creating two integers. Setting the first to the value 1 and then initializing the second variable by pointing it to the first. Writing out the initial values we see that both variables have the same value of 1. The next thing we do is change the value or i = 2. Then we write out the values of the variables and see that i = 2 but j = 1. The reason for this is that value types when assigned a value contain the actual data. So when we created j = i we got a copy of the actual data that i contained and j now has its own data. So when we come along later and make a change to i we are only changing its information and j will still reflect the information it holds.

// Begin Test
int i = 1;
int j = i;

Console.WriteLine(i); // Output = 1
Console.WriteLine(j); // Output = 1

i = 2;

Console.WriteLine(i); // Output = 2
Console.WriteLine(j); // Output = 1
// End Test

To be continued. Will be adding tests for passing into methods Structure (ValueType) and Class (ReferenceType)

 

From Cobalt to BlueQuartz and now BlueOnyx

It has been a bit of a journey since I first started offering web hosting services using the Cobalt Platform that was later sold to Sun Microsystems and then dropped from the product line. However thanks to Sun they released on the cobalt software as opensource and released to the public. Since then a great group of guys dedicated there time to producing an easy to install version based on CentOS called BlueQuartz.

Well it looks like it has now evolved to the next level. BlueOnyx is the latest attempt of keeping this opensource solution alive and going using the latest CentOS 5.2 release offering many new features. Below is some of the verbage from the BlueOnyx site http://www.blueonyx.it

--------------------------------------------------------

BlueOnyx is a Linux distribution (based on CentOS v5.2) which aims at delivering a turnkey server appliance for webhosting. It comes with a web based GUI interface which allows you (and your email, FTP and webhosting clients!) to manage most aspects of the server, its sites and accounts through a browser based interface. It is open source software, released under a Sun modified BSD license.

BlueOnyx is a modernized fork of BlueQuartz and has been developed by Solarspeed Ltd. and NuOnce Networks, Inc. and A lot of improvements and enhancements (compared to a regular BlueQuartz) went into BlueOnyx. Including support for Java Servlet Pages, built in web based email, phpMyAdmin for you and your webhosting clients and provisions to automatically create MySQL databases and users for each virtual site.

Visit BlueOnyx site for more information http://www.blueonyx.it

C# Traceroute using .net framework

If you have ever tried to create a traceroute program using one of the few available ICMP libraries freely available for C# you may have run into some issues mainly to do with the ICMP checksum not being correct. It seems that as of .NET 2.0 framework that microsoft included a Ping class that makes it really easy to then use it to create a traceroute utility. Here is some basic code to create a traceroute utility.

public string Traceroute(string ipAddressOrHostName)
{
    IPAddress ipAddress = Dns.GetHostEntry(ipAddressOrHostName).AddressList[0];
    StringBuilder traceResults = new StringBuilder();
 
    using(Ping pingSender = new Ping())
    {
        PingOptions pingOptions = new PingOptions();
        Stopwatch stopWatch = new Stopwatch();
        byte[] bytes = new byte[32];
 
        pingOptions.DontFragment = true;
        pingOptions.Ttl = 1;
        int maxHops = 30;
 
 
 
        traceResults.AppendLine(
            string.Format(
                "Tracing route to {0} over a maximum of {1} hops:", 
                ipAddress, 
                maxHops));
 
        traceResults.AppendLine();
 
        for(int i = 1; i < maxHops + 1; i++)
        {
            stopWatch.Reset();
            stopWatch.Start();
            PingReply pingReply = pingSender.Send(
                ipAddress, 
                5000, 
                new byte[32], pingOptions);
 
            stopWatch.Stop();
 
            traceResults.AppendLine(
                string.Format("{0}\t{1} ms\t{2}", 
                i, 
                stopWatch.ElapsedMilliseconds, 
                pingReply.Address));
 
            if(pingReply.Status == IPStatus.Success)
            {
                traceResults.AppendLine();
                traceResults.AppendLine("Trace complete."); break;
            }
 
            pingOptions.Ttl++;
        }
 
    }
 
    return traceResults.ToString();
}

 

 

Microsoft released free Chart Controls for ASP.NET and Winforms

I might be behind the times but I just found out that MS released free charting controls for Winform and ASP.NET. Requires you be running VS2008.

Charting Component Download : http://www.microsoft.com/downloads/details.aspx?FamilyId=130F7986-BF49-4FE5-9CA8-910AE6EA442C&displaylang=en

Language Pack if you are using other languages: http://www.microsoft.com/downloads/details.aspx?FamilyId=581FF4E3-749F-4454-A5E3-DE4C463143BD&displaylang=en

Visual Studio 2008 Add-On: http://www.microsoft.com/downloads/details.aspx?familyid=1D69CE13-E1E5-4315-825C-F14D33A303E9&displaylang=en

Documentation: http://www.microsoft.com/downloads/details.aspx?familyid=EE8F6F35-B087-4324-9DBA-6DD5E844FD9F&displaylang=en

Also here is a good resouce site that provides samples in asp.net and winforms for using the controls: http://code.msdn.microsoft.com/mschart/Release/ProjectReleases.aspx?ReleaseId=1591

I will be downloading later this evening and playing around with this but this is great to see MS finally release some nice looking chart controls.

c# and vb.net preserving or persisting user settings user.config after upgrades

If you store settings in your application using Settings.Settings and have ever upgraded your application you may have found all the user settings have vanished. It seems that the user.config file is stored in %UserProfile/LocalSettings/Application Data/CompanyName/ApplicationName/versionnumber/user.config

Every time you increment the version on your application a new version folder will be created that is of course empty and the process of populating those user specific settings starts over again.

Here is a way you can use to persist those settings after an upgrade.

C# Directions:

In your project open Settings.Settings and add a new user setting called ShouldUpgrade and set this to be a boolean value and set the value to true. Now click View Code icon at the top left side of the settings window. In the constructor add something like this.

public Settings()
{
    if(this.ShouldUpgrade)
    {
       
this.Upgrade();
       
this.ShouldUpgrade = false;
       
this.Save();
    }
}

So the first time that the settings class is instantiated it will check the ShouldUpgrade setting of the new user.config file which will be set to true. The this.Upgrade() then copies any user settings from the most recent version prior to this upgrade. Then set ShouldUpgrade to false and save the settings.

Also because we stored ShouldUpgrade as a user setting this upgrade will happen for each individual user on the same machine.

For VB.NET you do the same approach but use My.Settings.Upgrade()

 

C# Ping

I was recently tasked with putting together a ping and traceroute feature in an application and like I always do I googled to see if someone had already worked out how to do this.

I found a few people who had posted ICMP libraries and so I used one of them and tested and everything seemed to work properly. However I was getting reports that it was not working on Vista. However the person could use the built in ping and tracert utilities with no issue.

At first I thought it might be related to permissions or firewall but none of those ended up resolving the issue.

So I fired up wireshark and looked at what was different at the packet level between what I was doing and what the built in windows ping was doing.

The code I was using was returning an incorrect checksum on the ICMP reply. So I searched a bit to learn about the checksum and ended up finding somehow that as of .NET 2.0 framework there is a Ping class built into the framework!!

So I am posting this in hopes of saving someone a bunch of time with the same issue I had. Also the code I had to put together initially was much more complex. This I think you will see is very straight forward.

Download PingUtility.cs (3.14 kb)

Example Output:

Pinging 192.168.105.10 with 32 bytes of data:

Reply from 192.168.105.10: bytes=32 time=11ms TTL=128
Reply from 192.168.105.10: bytes=32 time=3ms TTL=128
Reply from 192.168.105.10: bytes=32 time=3ms TTL=128
Reply from 192.168.105.10: bytes=32 time=3ms TTL=128

Ping statistics for 192.168.105.10:
	Packets: Sent = 4, Received = 4, Lost = 0
Approximate round trip times in milli-seconds:
	Minimum = 3ms, Maximum = 11ms
Code:
public static string Ping()
{
    using (Ping pingSender = new Ping())
    {
        PingOptions pingOptions = null;
        StringBuilder pingResults = null;
        PingReply pingReply = null;
        IPAddress ipAddress = null;
        int numberOfPings = 4;
        int pingTimeout = 1000;
        int byteSize = 32;
        byte[] buffer = new byte[byteSize];
        int sentPings = 0;
        int receivedPings = 0;
        int lostPings = 0;
        long minPingResponse = 0;
        long maxPingResponse = 0;
        string ipAddressString = "192.168.105.10";
        pingOptions = new PingOptions();
        //pingOptions.DontFragment = true;
        //pingOptions.Ttl = 128;
        ipAddress = IPAddress.Parse(ipAddressString);
        pingResults = new StringBuilder();
        pingResults.AppendLine(string.Format("Pinging {0} with {1} bytes of data:", ipAddress, byteSize));
        pingResults.AppendLine();
        for (int i = 0; i < numberOfPings; i++)
        {
            sentPings++;
            pingReply = pingSender.Send(ipAddress, pingTimeout, buffer, pingOptions);
            if (pingReply.Status == IPStatus.Success)
            {
                pingResults.AppendLine(string.Format("Reply from {0}: bytes={1} time={2}ms TTL={3}", ipAddress, byteSize, pingReply.RoundtripTime, pingReply.Options.Ttl));
                if (minPingResponse == 0)
                {
                    minPingResponse = pingReply.RoundtripTime;
                    maxPingResponse = minPingResponse;
                }
                else if (pingReply.RoundtripTime < minPingResponse)
                {
                    minPingResponse = pingReply.RoundtripTime;
                }
                else if (pingReply.RoundtripTime > maxPingResponse)
                {
                    maxPingResponse = pingReply.RoundtripTime;
                }
                receivedPings++;
            }
            else
            {
                pingResults.AppendLine(pingReply.Status.ToString());
                lostPings++;
            }
        }
        pingResults.AppendLine();
        pingResults.AppendLine(string.Format("Ping statistics for {0}:", ipAddress));
        pingResults.AppendLine(string.Format("\tPackets: Sent = {0}, Received = {1}, Lost = {2}", sentPings, receivedPings, lostPings));
        pingResults.AppendLine("Approximate round trip times in milli-seconds:");
        pingResults.AppendLine(string.Format("\tMinimum = {0}ms, Maximum = {1}ms", minPingResponse, maxPingResponse));
        return pingResults.ToString();
    }
}


BlueQuartz not generating usage information for sites

For those that might be having the same problem here is what I found.

Going to Usage Information and clicking to generate web, mail or ftp stats resulted in no stats available for period. I looked for a recent site created and tried and stats worked. When I went and looked at what the difference was between the site that worked and the one did not I found there was a permission difference on the logs folder.

Site that worked
drwxr-s---  3 nobody site26  4096 Nov 13 03:28 logs

Site that did not work
drwxr-s--x   4 SITE66-logs site66  4096 Dec 11 12:30 logs

To fix this I issued this command to reset on all sites the permission on the logs folder

find /home -type d -name 'logs' -mindepth 4 -print | xargs chmod 02751

Now all my sites are able to generate stats again.