Method 1 involves recursion. I saw a sample at http://www.gfilter.net/default.aspx?Post=Code-Snippet-of-the-Day%3a-Recursive-File-Finder and MSDN.

Method 2 involves using a stack to process subdirectories in a Last in, first out manner.  I saw this from http://dotnetperls.com/Content/Recursively-Find-Files.aspx.

The result from finding all 13,300+ DLLs in c:\windows and all subdirectories:

FileFinderRecursive Elapsed: 6876ms.
FileFinderStack Elapsed: 5725ms.

The stack method was almost 20% faster.

Here’s some code from a quick C# console app I just wrote to compare the two:

class Program
{
    static void Main(string[] args)
    {
        Stopwatch s;
        IFinder finder;
        string startDir = @"c:\windows";
        string matchType = "*.dll";

        //Test Recursion method
        s = Stopwatch.StartNew();
        finder = new FileFinderRecursive();
        var filesRecursive = finder.Find(startDir, matchType);
        s.Stop();
        Console.WriteLine("FileFinderRecursive Elapsed: " + s.ElapsedMilliseconds + "ms.");

        //Test Stack method
        s = Stopwatch.StartNew();
        finder = new FileFinderStack();
        var filesStack = finder.Find(startDir, matchType);
        s.Stop();
        Console.WriteLine("FileFinderStack Elapsed: " + s.ElapsedMilliseconds + "ms.");    

        Console.Read();
    }
}
public interface IFinder
{
    List<FileInfo> Find(string startDir, string matchType);
}

/// <summary>
/// Recursively finds a list of files that match the specified critiria
/// </summary>
public class FileFinderRecursive : IFinder
{
    private Dictionary<FileInfo, int> files;

    public List<FileInfo> Find(string startDir, string matchType)
    {
        files = new Dictionary<FileInfo, int>();
        findFiles(startDir, matchType);

        return files.Keys.ToList();
    }

    private void findFiles(string startDir, string matchType)
    {
        DirectoryInfo d = new DirectoryInfo(startDir);
        FileInfo[] fileInfos;
        try
        {
            fileInfos = d.GetFiles(matchType);
        }
        catch
        {
            return; //most likely Access Exception.  BAIL!
        }
        foreach (FileInfo f in fileInfos)
        {
            if (!files.ContainsKey(f))
            {
                files.Add(f, 0);
            }
        }

        if (d.GetDirectories().Length > 0)
        {
            foreach (DirectoryInfo subD in d.GetDirectories())
            {
                findFiles(subD.FullName, matchType);
            }
        }
    }
}

/// <summary>
/// Finds a list of files that match the specified criteria using a stack
/// </summary>
public class FileFinderStack : IFinder
{
    public List<FileInfo> Find(string startDir, string matchType)
    {
        DirectoryInfo dirStart = new DirectoryInfo(startDir);
        List<FileInfo> result = new List<FileInfo>();
        Stack<DirectoryInfo> stack = new Stack<DirectoryInfo>();
        stack.Push(dirStart);
        while (stack.Count > 0)
        {
            DirectoryInfo dir = stack.Pop();
            try
            {
                result.AddRange(dir.GetFiles(matchType));
                DirectoryInfo[] subDirs = dir.GetDirectories();
                foreach (DirectoryInfo dn in subDirs)
                {
                    stack.Push(dn);
                }
            }
            catch { }
        }
        return result;
    }
}

a

I’ve rewritten his JavaScript code here in C# with a few guard clauses at the top. Enjoy!
================================

private static bool ValidateSIN(string sin)
{
    int dummy;
    if (!Int32.TryParse(sin, out dummy)) return false;
    if (sin.Length != 9) return false;

    int s = 0;
    for (int i = 0; i < 9; i++)
    {
        int x = Convert.ToInt32(sin.Substring(i, 1));
        if (i % 2 != 0)
        {
            if ((x << 1) > 9)
            {
                s += ((x << 1) - 9);
            }
            else
            {
                s += (x << 1);
            }
        }
        else
        {
            s += x;
        }
    }
    return (s % 10 == 0);
}

EDIT: Neil correctly pointed out I was missing a null check as my first statement; I was checking (sin.Length != 9) first.  I’ve changed it and decided to do the Int32.TryParse() first before checking for length.  According to Reflector, System.Int32.TryParse(string s, out int result) calls the internal method System.Number.TryParseInt32(s, NumberStyles.Integer, NumberFormatInfo.CurrentInfo, out result) which in turn calls System.Number.TryStringToNumber(string str, NumberStyles options, ref NumberBuffer number, NumberFormatInfo numfmt, bool parseDecimal) which does a null check.

I like Rick Brewster’s fluent approach to parameter validation using .NET 3.0 and higher.  It’s pretty slick.

a

The DLink DNS-323 is a network-attached storage enclosure that holds 2 SATA drives.  I decided to buy 2 1-terabyte Seagate Barracuda 7200.11 drives which were also on sale.

For the past little while, we’ve been mostly using stand-alone drives with USB enclosures.  Some of them we’ve bought as a retail package.  Others, we’ve bought the drive first and then slapped an enclosure around it.  But all are USB-attached and provide no fault tolerance.

Herein lies the appeal of the DNS-323.  This unit can take 2 SATA drives and can use them in the following configurations:

• Normal – The 2 disks are 2 individual drives
• JBOD – aka “Just a Bunch of disks”
• RAID 0 – Striped
• RAID 1 – Mirrored

For our purposes, we will be storing some valuable data (pics, videos, etc.) on this unit so reliability, redundancy and fault-tolerance is top priority.  I opted, therefore, to configure this in RAID 1.

Installation was a breeze – install hard drives, connect to router and power up.  I did have a problem with my VMWare Server’s built-in DHCP server assigning the DNS-323 an IP address.  The DNS-323 was supposed to get its IP address from my Linksys WRT54G router instead.  Luckily, the DNS-323’s included software has a tool to find your unit on the network.  Once I found the unit on the network, I accessed its web-based configuration interface and assigned a static IP address.

The DNS-323 has a built-in FTP server, iTunes server, BitTorrent client and also acts as a UPnP AV media server.  The DNS-323 runs Linux and is accessible via Telnet.  For more info on the DNS-323, you can visit the Wiki.

The bottleneck with our setup will most likely be our network until we can upgrade our router to Gigabit Ethernet.  My aging WRT54G is only a 100 Mbps router.  So transferring files from my computer to the DNS-323 will run at 12.5 megabytes per second at the fastest.  It will most likely be around 10 MB/s.

a

I used to visit DoFactory’s patterns pages for all the patterns but now I have a sheet with all of them on a page.  Cool!

a

So why do I always see sample code of an integer addition method written in C# like this:

private static int add(int a, int b){
    return a + b;
}

So if I were to add 2,000,000,000 and 2,000,000,000 – an integer datatype would overflow. But the solution isn’t merely to set the return value to a “long” (i.e. 64-bit integer):

private static long add(int a, int b){
    return a + b;
}

You have to cast each parameter variable to a long before performing the addition:

private static long add(int a, int b){
    return (long)a + (long)b;
}

Here are some I found:

http://www.cs.tufts.edu/comp/194NET/notes/csharp.php3

http://www.harding.edu/fmccown/java1_5_csharp_comparison.html

http://www.developernotes.com/post/Use-Ruby-to-Unit-Test-C.aspx

a