The Blog of Josh Stodola

Sorting a Generic List Dynamically in VB.NET

Josh Stodola - josh420@josh420.com — Tue, 03 Feb 2009 14:44:27 -0600

In a recent post, I talked about generic lists and how we have been using them at work as a replacement for the DataTable class. For the most part, it has been a real pleasure getting rid of DataTable, but I will admit that there are certain situations in which a DataTable is easier to code for. This is primarily because DataTables are mature and robust, and several helper classes already exist in the .NET framework to process them. The Fill function for the SqlDataAdapter, for example, will automatically load a DataTable with the result of an SQL command for you, which is extremely convenient. There are other classes such as DataView that make manipulative tasks (like sorting and filtering) a piece of cake. For instance, say you have a DataTable populated with about 500 rows and you need to re-sort it. This can be achieved very easily with the DataView class like this…

Private Function SortDataTable(ByRef Table as DataTable, ByVal SortString as String) as DataTable
    Dim View as New DataView(Table)
    View.Sort = SortString
    Return View.ToTable()
End Function

Man, that's almost too easy! It is capable of sorting by multiple fields in any order using a very simple syntax similiar to the "ORDER BY" clause in SQL. As you could probably guess, it is more difficult to sort a generic list. Thanks to the inherent flexibility of generics and reflection, however, it is possible to achieve identical functionality. And I've already done the work to make this happen. The class I created uses reflection, which does not perform fantastically, but it offers up unprecedented flexibility. If performance/efficiency is of noteworthy concern in your application, you might want to consider using a type-specific approach. Also, it was mentioned on Stack Overflow that C# developers can use inline delegates and VB9 developers can use lambda expressions, which I thought was an advanced yet elegant approach. For the sake of simplicity, I just created an instance class that inherits IComparer and will work with any type on the .NET 2.0 framework...

Imports System.Collections
Imports System.Collections.Generic
Imports System.Reflection

Namespace Utility

    ''' <summary>
    ''' Sort order enumeration
    ''' </summary>
    Public Enum SortOrder
        Ascending
        Descending
    End Enum

    ''' <summary>
    ''' This instance class is used to sort a generic collection of object instances.
    ''' It automatically fetches the type and performs the necessary comparison(s) to sort.
    ''' 
    ''' To use, instantiate this class, set the sort string property, and pass this
    ''' instance to the internal Sort() function of your generic collection.
    ''' 
    ''' Example:
    '''     Dim MyList As List(Of MyClassType) = 'Populate the list somehow
    '''     Dim Sorter As New Sorter(Of MyClassType)
    '''     Sorter.SortString = "Field1 DESC, Field2"
    '''     MyList.Sort(Sorter) 'After this call, the list is sorted
    ''' </summary>
    Public Class Sorter(Of T)
        Implements IComparer(Of T)

        Private _Sort As String

        ''' <summary>
        ''' Instantiate the class.
        ''' </summary>
        Public Sub New()

        End Sub

        ''' <summary>
        ''' Instantiate the class, setting the sort string.
        ''' 
        ''' Example: "LastName DESC, FirstName"
        ''' </summary>
        Public Sub New(ByVal SortString As String)
            _Sort = SortString
        End Sub

        ''' <summary>
        ''' The sort string used to perform the sort. Can sort on multiple fields.
        ''' Use the property names of the class and basic SQL Syntax.
        ''' 
        ''' Example: "LastName DESC, FirstName"
        ''' </summary>
        Public Property SortString() As String
            Get
                If _Sort IsNot Nothing Then
                    Return _Sort.Trim()
                End If

                Return Nothing
            End Get
            Set(ByVal value As String)
                _Sort = value
            End Set
        End Property

        ''' <summary>
        ''' This is an implementation of IComparer(Of T).Compare
        ''' Can sort on multiple fields, or just one.
        ''' </summary>
        Public Function Compare(ByVal x As T, ByVal y As T) As Integer Implements IComparer(Of T).Compare
            If Not String.IsNullOrEmpty(Me.SortString) Then
                Const ERR As String = "The property ""{0}"" does not exist in type ""{1}"""
                Dim Type As Type = GetType(T)
                Dim Comp As Comparer = Comparer.DefaultInvariant
                Dim Info As PropertyInfo

                For Each Expr As String In Me.SortString.Split(","c)
                    Dim Dir As SortOrder = SortOrder.Ascending
                    Dim Field As String

                    Expr = Expr.Trim()

                    If Expr.EndsWith(" DESC") Then
                        Field = Expr.Replace(" DESC", String.Empty).Trim()
                        Dir = SortOrder.Descending
                    Else
                        Field = Expr.Replace(" ASC", String.Empty).Trim()
                    End If

                    Info = Type.GetProperty(Field)

                    If Info Is Nothing Then
                        Throw New MissingFieldException(String.Format(ERR, Field, Type.ToString()))
                    Else
                        Dim Result As Integer = Comp.Compare(Info.GetValue(x, Nothing), Info.GetValue(y, Nothing))

                        If Result <> 0 Then
                            If Dir = SortOrder.Descending Then
                                Return Result * -1
                            Else
                                Return Result
                            End If
                        End If
                    End If
                Next
            End If

            Return 0
        End Function
    End Class
End Namespace

The meat of the class lies in the Compare function, which is the essential drivetrain of the IComparer interface. If you are not at all familiar with IComparer, you should check out this article on C# corner about comparing objects; it's quite a useful concept to grasp. The class is fully capable of sorting on one or multiple fields of any type of object. The "sort string" uses the same basic syntax as the ORDER BY clause in SQL, and the field names to use are actually the property names within your class. All you have to do is instantiate the Sorter class, set the type and the sort string, and then pass the class instance to the Sort() function of your generic collection. Following is a simple example (called ConsoleApplication149, for me!) that fully demonstrates the flexibility...

Module Module1

    Sub Main()
        Dim People As List(Of Person) = GetPeople()
        Dim Sorter As New Sorter(Of Person)

        Sorter.SortString = "FirstName"
        People.Sort(Sorter)
        PrintResults(People, Sorter.SortString)

        Sorter.SortString = "LastName DESC, FirstName, MiddleName"
        People.Sort(Sorter)
        PrintResults(People, Sorter.SortString)

        Sorter.SortString = "DateOfBirth DESC"
        People.Sort(Sorter)
        PrintResults(People, Sorter.SortString)

        Console.ReadLine()
    End Sub

    Function GetPeople() As List(Of Person)
        Dim Result As New List(Of Person)

        With Result
            .Add(New Person("John", "M", "Doe", #1/1/1969#))
            .Add(New Person("Jane", "A", "Doe", #3/4/1972#))
            .Add(New Person("Paul", "L", "Smith", #2/1/1948#))
            .Add(New Person("Janet", "A", "Doe", #9/16/1975#))
            .Add(New Person("Patricia", "B", "Smith", #3/14/1952#))
            .Add(New Person("John", "Matthew", "Doe", #12/21/1988#))
            .Add(New Person("James", "L", "Doe", #6/19/1990#))
            .Add(New Person("Patrick", "O", "Smith", #8/26/1993#))
        End With

        Return Result
    End Function

    Sub PrintResults(ByRef People As List(Of Person), ByVal SortString As String)
        Console.WriteLine("Sort String: " & SortString)
        Console.WriteLine()

        For Each Per As Person In People
            Console.WriteLine(Per.ToString())
        Next

        Console.WriteLine()
        Console.WriteLine()
    End Sub
End Module

Class Person

    Private _First As String
    Private _Middle As String
    Private _Last As String
    Private _Dob As Date

    Sub New()

    End Sub

    Sub New(ByVal FirstName As String, ByVal MiddleName As String, ByVal LastName As String, ByVal DateOfBirth As Date)
        _First = FirstName
        _Middle = MiddleName
        _Last = LastName
        _Dob = DateOfBirth
    End Sub

    Property FirstName() As String
        Get
            Return _First
        End Get
        Set(ByVal value As String)
            _First = value
        End Set
    End Property

    Property MiddleName() As String
        Get
            Return _Middle
        End Get
        Set(ByVal value As String)
            _Middle = value
        End Set
    End Property

    Property LastName() As String
        Get
            Return _Last
        End Get
        Set(ByVal value As String)
            _Last = value
        End Set
    End Property

    Property DateOfBirth() As Date
        Get
            Return _Dob
        End Get
        Set(ByVal value As Date)
            _Dob = value
        End Set
    End Property

    Overrides Function ToString() As String
        Return Me.FirstName.PadRight(15, " ") & _
               Me.MiddleName.PadRight(10, " ") & _
               Me.LastName.PadRight(15, " ") & _
               Me.DateOfBirth.ToString("yyyy-MM-dd")
    End Function
End Class

Simple as that! I've grown to love this class, and I hope you too find it useful!

Javascript Learning Pains - parseInt does not work!

Josh Stodola - josh420@josh420.com — Fri, 30 Jan 2009 09:42:35 -0600

Javascript is a very unique language, and I think that has become more and more apparent over the years. Crockford said it best - it's the worlds most misunderstood programming language. And yet it has somehow managed to evolve into a sheer necessity of the web. I can blatantly recall my very first programming experience with Javascript, back in 1998. I spent over an hour trying to figure out why my one line of code to get a form element was not working. Then I found out that Javscript was a case-sensitive language, and this really pissed me off. I could use whatever case I wanted in HTML, so why does Javascript have to be any different?! Regardless of your level of experience as a programmer (in general), you should probably expect some introductory hiccups upon first learning Javascript. I believe that such initial frustrations are why the language initially got such a bad rap. To become an expert, you have to be persistent, inquisitive, and down-right stubborn.

The other day, a co-worker of mine was flipping his wig over the parseInt() function. It was not working consistently, which was probably pretty aggravating. Hell, I know it was aggravating. He came to the realization that it was failing everytime the given number had a leading zero. No, this is not a bug. The problem (of course!) was that he was not specifying a radix. Yes, parseInt() has two parameters, not just one. When you do not specify a radix, it makes an educated guess at what it should be, based on a simple evaluation of the input string. W3Schools clearly states…

If the string begins with "0x", the radix is 16 (hex)
If the string begins with "0", the radix is 8 (octal)
Otherwise, the radix is 10 (decimal)

That said, passing a value of "080" without specifiying a radix is always going to return zero because it is not a valid Octal value. Go ahead, see for yourself. The fix is really simple, and becomes painfully obvious once you understand the above. Make sure to always specify a decimal radix, and it will work. People who have no patience for specifications might realize that using parseFloat instead of parseInt is also a potential workaround. But if you are like me, you have to know why something does not work. And this is just one of the few oddball issues you can expect to encounter if you are new to Javascript.

If you have managed to master the Javascript language without having at least one aggravating encounter like this, you must be inhuman!

Creating an Excel Spreadsheet from a Generic List in VB.NET

Josh Stodola - josh420@josh420.com — Mon, 26 Jan 2009 18:37:06 -0600

I'm a huge fan of generic lists. They have numerous benefits, and because of that I've been campaigning them at work to the other programmers. The response has been very good. Although given that our previous method was powered by DataTables, it is not much of a surprise that generic lists have evolved into the preference. A question came up today about adding the capability to export a GridView's data source to a Microsoft Excel spreadsheet, and it just so happens the GridView was bound to a List. This is the first time this has come up. We've had several grids export to Excel, and a common function to handle most of the workload was in place, but unfortuntely it was built for a DataTable instance. This task was accomplished fairly simply with DataTables, because it was unnecessary to worry about the parameter type - it was always a DataTable. The function loops through all the rows and builds a tab/line delimited string of data and returns it. DataTables also made it simple to get the name of the columns (which may be used as column headers). Thanks to generics and reflection, I was able to build a function that performs the same task with equivalent flexibility…

''' <summary>
''' Returns a string that can be streamed out to an Excel spreadsheet.
''' Records are delimited by a line feed (13). Fields are delimited by a tab (9).
''' </summary>
''' <param name="Objects">The list of class instances to convert</param>
''' <param name="IncludeHeadings">
''' Include a heading record at the beginning of the string? 
''' Headings will be derived from the property name.
''' </param>
''' <param name="IncludeReadOnlyProperties">
''' Include ReadOnly properties in the response for each instance?
''' </param>
''' <returns>Nothing if the given collection is empty</returns>
Public Shared Function ListToString(Of T)(ByVal Objects As List(Of T), ByVal IncludeHeadings As Boolean, ByVal IncludeReadOnlyProperties As Boolean) As String
    If Objects Is Nothing OrElse Objects.Count = 0 Then
        Return Nothing
    End If

    Dim TAB As String = Convert.ToChar(9)
    Dim LF As String = Convert.ToChar(13)
    Dim Result As New StringBuilder()
    Dim Type As Type = GetType(T)
    Dim Props As PropertyInfo() = Type.GetProperties()

    If IncludeHeadings Then
        For Each Prop As PropertyInfo In Props
            If Prop.CanWrite OrElse IncludeReadOnlyProperties Then
                Result.Append(Prop.Name)
                Result.Append(TAB)
            End If
        Next
    End If

    For Each Obj As Object In Objects
        Result.Append(LF)

        For Each Prop As PropertyInfo In Props
            If Prop.CanWrite OrElse IncludeReadOnlyProperties Then
                Result.Append(Prop.GetValue(Obj, Nothing).ToString())
                Result.Append(TAB)
            End If
        Next
    Next

    Return Result.ToString()
End Function

''' <summary>
''' Returns a string that can be streamed out to an Excel spreadsheet.
''' Records are delimited by a line feed (13). Fields are delimited by a tab (9).
''' Includes headings. Excludes read-only properties.
''' </summary>
''' <param name="Objects">The list of class instances to convert</param>
''' <returns>Nothing if the given collection is empty</returns>
Public Shared Function ListToString(Of T)(ByVal Objects As List(Of T)) As String
    Return ListToString(Objects, True, False)
End Function

And of course, I should mention that Reflection in the .NET framework is not the most efficient approach in the world. That said, if you are setting up something to process batch, you might want to consider using a strongly-typed method. However, if the intended functionality is not expected to be used frequently, then this should work just fine for you. You have to make an educated decision on whether or not the efficiency impact is crucial enough to supercede the convenience factor of having a global function. In our case, it certainly was not.

Also, I want to make it clear that I am well-aware of the approach to exporting GridViews to excel that renders the grid's HTML and relies on Excel to interpret it correctly. So there is no need to point that out to me (although somebody probably will anyways). I am not really a fan of this method. It has it's purpose and it is undeniably simplistic, but it was not right for me. If you are in a similiar situation (you need to get at the data source of the grid, or you aren't even using a visual grid), you'll appreciate the function above.

Here's a quick and dirty example of how to use the function...

Structure Employee
    Private _Id As Integer
    Private _Name As String
    Private _DateHired As DateTime

    Sub New(ByVal Id As Integer, ByVal Name As String, ByVal DateHired As DateTime)
        _Id = Id
        _Name = Name
        _DateHired = DateHired
    End Sub

    ReadOnly Property Id() As Integer
        Get
            Return _Id
        End Get
    End Property

    Property Name() As String
        Get
            Return _Name
        End Get
        Set(ByVal value As String)
            _Name = value
        End Set
    End Property

    Property DateHired() As DateTime
        Get
            Return _DateHired
        End Get
        Set(ByVal value As DateTime)
            _DateHired = value
        End Set
    End Property
End Structure

Sub Main()
    Dim MyList As New List(Of Employee)
    Dim RightNow As DateTime = DateTime.Now
    Dim Export As String

    MyList.Add(New Employee(1, "Josh Stodola", RightNow.AddDays(-365)))
    MyList.Add(New Employee(2, "John Doe", RightNow.AddDays(-31)))
    MyList.Add(New Employee(3, "Jane Doe", RightNow))

    Export = Utility.ListToString(MyList)

    Console.WriteLine()
    Console.WriteLine(Export)
    Console.WriteLine()
    Console.ReadLine()
End Sub

Hope you found this useful! Oh, and accept my apologies for the inexcusable lack of blog posts lately.

Third Week of August

Josh Stodola - josh420@josh420.com — Fri, 15 Aug 2008 12:19:10 -0600

It's finally that time of year again. The time to shut down the computers. The time to relax and put the mind at ease. The time to pigeonhole the daily tasks that are seemingly important, and realize that they are actually humdrum. The time to part myself from this electronics-infatuated lifestyle, in an effort to maintain natural sanity.

It's the third week of August, and that means one thing to me: fishing time. Tomorrow morning, bright and early, I will be on my way to Minnesota again to spend a computer-free week with nature, with the mere objective of chillaxing. This is an annual vacation for me, and I could almost classify such a trip as essential for anybody who uses computers for several hours a day (given that you are not disturbed enough to say you'd have sex with computers). Escaping the shackles that are the keyboard and mouse, even for a day or two, has a profound effect on my general outlook towards life. Spending 70 hours of the week behind a computer is somewhat humbling, even if you have a sincere passion for it. I certainly do not want to become one who thouroughly enjoys doing something so much, but does it so often that he/she ends up growing tired of it.

To combat that nightmare of a possibility, I just get out on a beautiful lake on a beautiful day and forget about everything (except my way home of course). And it does the trick for me. The best coding day of the year is always the first day after I return from vacation. It's like a weird sort of revelation. So if you are one who works behind a computer for a living, and then retires to the household to spend more time on the computer, please do yourself a favor and make an effort to escape on a yearly basis (at least). And do not bring your laptop with you!

As a result of my absence, replies to email and comments will be delayed until August 24th.

My New Office

Josh Stodola - josh420@josh420.com — Wed, 30 Jul 2008 21:12:17 -0600

I've been saving up money for a little while now, with hopes of revamping my office at home to accomodate my ever-growing computing ~~needs~~ wants. I was finally able to purchase all that I wanted, and now I am quite happy with my home within my home. It's amazing how dropping a few grand on yourself can change your whole outlook on life. The ultimate addition to my office is obviously a new computer, but before I could go down that route I decided it would be best to get a new desk first. I'm sold on the multiple monitors concept, so I needed a spacious desk with room for at least two big LCDs for the new machine. Also, my old computer (ha, I can call it old now) still has plenty of life left, so I needed a desk that could accomodate it as well, without being cluttered. After shopping at a multitude of stores around town, I threw in the towel on the traditional market and turned to the web. I should've done that to begin with; I wasted too much time and gas wandering aimlessly, and I had to put up with a bunch of annoying salesmen. I don't need to be persuaded into believing that L-shaped desks are far superior to anything else! Damn salesmen. Anyways, I ended up settling on a desk from VersaTables.

I Heart VersaTables

They specialize in computer furniture for classrooms, computer labs, and medical establishments. They have adjustable desks, wall mount setups, carts, drafting tables, etc. No, I am not setting up a computer lab here, but I like these styles of desks because they are built to last. I purchased the 72" Encosed Classroom Desk, and I think it is fantastic. The first thing I noticed about this desk is its level of sturdiness. I could treat it like a trampoline and it would not budge. Also, they probably could've just dropped it right from the FedEx airplane, because it was packed better than anything I've ever seen. Not a single chip or scratch in the black paint, I was very impressed. You might be wondering, where are the drawers? It has none. I had a desk with drawers before, and I never used them. My life is practically paperless, so I really don't have a need for file cabinets and drawers; they just get in the way. In my eyes, the lack of drawers was actually a selling point. The desk comes with a nice cable management tray in the back, which helps you organize your coords and reduce the amount of exposure. I accessorized this beast by getting two enclosed CPU holders, an adjustable keyboard arm and tray for my new PC, and a keyboard drawer for the other machine. With these accessories, I have the whole desk surface to work with (aside from the space required for the monitors and speakers, of course). It's worth mentioning that the keyboard arm I purchased was not very sturdy, and it did not slide as smoothly as it should. I simply called them and talked about it and they were happy to send me a new one free of charge (and they spoke to me like a person, in clear English). Now it works perfectly. They didn't even ask me to ship the other one back. Now that is customer service. These desks are made in the USA and come with a lifetime warranty. I could not be happier.

My Buttocks Rejoice

Unfortunately, I was completely unable to find any Herman-Miller dealers around here that would let me take one of their prestine chairs for a test-trial, so I threw the Aeron and Mirra options right out the window. I've heard great things about them, but I guess they should do a better job of marketing; I had to make an effort just to locate a dealer, and I still was unable to reach a human being. I ended up purchasing an all-black Steelcase Leap off eBay for roughly $450, and so far I am pleased with it. Although I do wish it had a little more support for the upper back. I am 6'4" so perhaps I need more support in this area than the average user. One remarkable thing about this chair is that it reclines very comfortably. I never thought leaning back could feel so good! In other chairs that I have sat in, leaning back tends to increase the pressure on my spine. The Leap does a good job of distributing the pressure evenly with the so-called "Natural Glide System". There is an ungodly amount of adjustments on this chair. That said, you can expect it to take some time to find your "spot", and don't dare let anybody else sit in it!

The Ultimate PC

I've always wanted to build my own PC. Not only do you get a bigger bang for your buck, but there is a certain sense of pride that comes along with it. This was my first build, and I will probably never turn back to retail. As a control-freak, I just love being able to say "Yup, I built this". Ever since I read the post from the coding whore himself, Jeff Atwood, I was intrigued to give it a shot. And, as planned, I tried to go all-out. This is what my new machine consists of…

Antec Sonata III Black Case - $124.95
This is a very nice and quiet case, although I am not so sure that opening a door to reach my DVD drives is desired. It doesn't really bother me, but I could see how it could be a major annoyance to some. Also, I think it could be ventilated better. Came with a really good 500W power supply.
GIGABYTE GA-X48-DS4 LGA 775 Intel X48 ATX Motherboard - $224.95
This was probably more of a motherboard than I needed, but it's very nice indeed. I think the documentation could have been better.
Intel Core 2 Quad Q6700 Kentsfield 2.66Ghz Quad-Core Processor - $274.99
This CPU is incredible. I considered getting an Extreme, but decided it was no way worth the price. I haven't overclocked this CPU yet (too chicken), but I'll bet it can handle it well. I plan to OC in the future, perhaps when I grow some balls.
MSI NX8800GT 512MB OC GeForce 256-bit GDDR3 Video Card - $169.99
This is an impressive video card, it handles my dual monitors like a champ. Such a work horse. Loaded with features I will probably never use. Overclocked to 660 MHz, this thing flies at decent temperatures. It's big, and it packs a punch.
(2) mushkin 4GB (2x2GB) DDR2 SDRAM 800 Dual Channel - $217.98
This is 8GB of high quality RAM. Extremely fast. Came with an exceptional $25 mail-in rebate, too.
Western Digital VelociRaptor 300GB 10,000 RPM Hard Drive - $299.99
This is the boot drive, and I cannot get over how fast and how quiet this hard drive is. It is literally sickening. Expensive, but I truly think it is worth it. Atwood even recommended it to RobCon. Get it. Now.
Western Digital Caviar 500GB 7,200 Hard Drive - $79.99
This is my secondary drive that I use primarily for multimedia storage. I just might buy a couple more of these, they are extremely cheap and would work great for backups. Could be quieter, of course, but the Antec case comes with these fantastic rubber grommets to help eliminate vibration noise.
Zalman CNPS9700 2 Ball CPU Cooler - $53.99
Supposed to be an excellent cooler, but I expected more. After 4 hours of Prime95, Core 0 had a temperature of 65C. I expected much lower than that; I'm not even overclocked!
(2) Samsung 22X DVD Burner SATA - $57.98
They work great, but I wish they were quieter when burning. Go ahead, call me a noise freak.
(2) Acer Black 22" Widescreen LCD Monitor - $439.98
A great deal, has excellent picture quality. I did have one dead pixel in each of them, but they are barely noticeable, only when the entire screen is black. I didn't notice them for 2 days.
Windows XP 64-bit - $139.99
Since I wholeheartedly feel that Vista is the biggest heap of shit Microsoft has released since Bob, I had to go with good ol' XP. And let me tell you, this OS is very fast. Excitingly fast. Thus far, I have had no problems whatsoever. It runs intensive programs (such as Visual Studio and Photoshop) like a true champ. One complaint: it's too fast. I find the instantaneous appearance of windows somewhat disorienting. I should get used to it :)

I also got a Microsoft ComfortCurve Keyboard, a 5-Button Wired IntelliMouse, and Logitech Speakers (I already have a quality Plantronics headset). In the end, I spent about $2,250 on this machine. Note: these prices are as of early July 2008. And that, folks, is my ultimate PC! I love it, and I am guessing it will last me a good five years, at least.

I've never made a post like this before, I hope you enjoyed it!

Using Sprite Images with for a Hover Effect

Josh Stodola - josh420@josh420.com — Tue, 29 Jul 2008 22:45:46 -0600

I am a huge fan of CSS sprites, a technique used to create image-based hover effects on a web page. It's clever and very efficient because both "image states" are contained in a single image, eliminating the need to make two requests to the server. It tends to go hand-in-hand with image replacement to create an appealing effect while retaining accessibility and search engine optimization. Having to use background images sort of limits the flexibility to an extent. I realized this when I wanted to use sprites with the <input type="image" /> tag for submitting a form. Since this tag sets the path to the image as an attribute, the classic method of applying sprites with background images cannot be used. The same challenge is presented with the <img> tag. There are alternative routes, such as using two separate images and wiring up some Javascript to pre-load the hover image and change the SRC attribute of the input tag onmouseover. But, since I love sprites so much, I wanted to try and make it happen that way. It's not overly difficult nor complex, but I thought I would make a post out of it anyways. For an example, I will be using the following image as the sprite…

The plan is to make the top half of the image visible by default. When the mouse hovers over it, the bottom half should show (this could be vice-versa very easily). I'm assuming you understand the concept behind sprites, so I won't waste any more time explaining their structure and benefits. It's worth mentioning that the dimensions of the image above ar 200x80. To make this work, an extra <div> will be needed…

<div id="submitButton">
  <input type="image" src="submit.png" alt="Submit Form" />
</div>

Since the <div> is a container for the image, we can set a specific height and hide the overflow so that the default state of our image (the top) is the only thing showing…

#submitButton {
  cursor:pointer;       /* Give it the hand cursor, like a link */
  height:40px;          /* Image has a height of 80px, only show the first half */
  overflow:hidden;      /* Hide the overflow */
  width:200px;          /* Width of the image */
}

After that, setting the image to display the other half when moused over is as easy as adding a negative top-margin on the :hover pseudo-element of the <div>…

#submitButton:hover input {
  margin-top:-40px;    /* Negative height of half the sprite, to push the image up */
}

And that is all there is to it! I have tried this in IE7, Firefox 2, Opera 9, and Safari and it worked just fine. IE6 did not produce a hover effect, but it degrades gracefully by just showing the top half. If you discover a browser that causes abnormal results, please enlighten me and the future readers of this post by posting a comment. I've setup a live demo and a ZIP file for you to download.

Have fun!

The Cookie-Cutter Image Hover Effect

Josh Stodola - josh420@josh420.com — Mon, 30 Jun 2008 22:49:12 -0600

Today I am going to share an interesting and super-cool way to create a dynamic hover effect for images. I call this flexible format the "cookie-cutter effect", and you'll find out why in a moment. First, I shall go off on a tangent to maintain my reputation. This day in age, alot of navigational bars on web sites are image-based. Designers prefer to use graphics because it helps make the navigation more stunning and visually appealing (eye candy). Another thing that is becoming nearly mandatory is a hover effect for each of the links, meaning the image changes slightly when the user has positioned their cursor over it. This sends an instant visual cue to the user and lets them know exactly what they are pointing at. It may seem a little silly to think a user does not know where their mouse is pointed, but that certainly is not the case. The map in the center of your local shopping mall with a big arrow that says "You are here" is not silly either, it is actually helpful to some. Unfortunately, there is a pretty durastic downfall to using graphics for key elements such as navigation: they seriously lack extensibility. Say you want to increase the brightness of the hover effect by a few shades. This task, at the very least, requires knowledge of a graphics editing program. Usually some design talent is required too. In many cases this situation cannot be avoided, but the approach that I am discussing in this post will at the very least allow greater flexibility than the norm.

Let's Begin

To demonstrate the technique, I went ahead and created some random image with my first name next to an odd shape. I stuck the shape in there to illustrate that this approach is applicable to more than just textual images. In the end, the main part of this image (currently white) will be able to be changed to any color we want using CSS only. Consider the base image to the right as the opposite of what we want. The plan is to turn the text into the transparent part, and the background of the the image to a solid white. The result will be somewhat of a template, if you will. We will be able to use any background color or image behind it to essentially alter the part of the image that is actually intended to represent the foreground. I became aware of the usefulness of this concept when I read Jason Santa Maria's post the other day about his new blogging expedition. He intends for each post to have its own design theme/color scheme. Currently, he has the inverse effect of what I am demonstrating in this post; the background color of his navigational links change. This was accomplished by using normal transparent images like the one above, and is still flexible and cool. But, what if you want the color of the text to change? I'm sure Jason knows the answer.

Cut It Out

And this is where the term cookie-cutter is derived. When you cut shapes out of the dough, you are left with with a transparent version of the shape (commonly referred to as a cutout). You could slide a piece of regular paper underneath it to make the shape display as white. You could quickly and easily switch to a yellow paper and change the color of the entire shape. You could even slide a photo of yourself under it! This "theory" offers up quite a bit of flexibility, and I am going show you how to bring it to the web. I guess the toughest part of replicating this concept is the actual creation the image. I will be using Adobe Photoshop CS2 to do it. This amazing piece of software makes the task much easier to accomplish than I originally anticipated.

Photoshop Time

I am a Photoshop newbie, so this should not be difficult to follow. To start off, here is a screenshot of my Photoshop window to show you where I am at (I added a temporary black background so the text is more legible)…

First, merge the text and shape layers together…

Execute the Select All command by pressing Ctrl + A. Then select the Move Tool from the toolbox. By pressing any of the arrow buttons on your keyboard, Photoshop will automatically select the contents of the layer (does anybody know an easier way to do this?)...

At this time we want the inverse of the current selection – the transparent part. This can be accomplished easily by using the keyboard shortcut Ctrl + Shift + I. Here is the result...

Now we shall fill the transparent selection with a solid white color. You could use a texture or a gradient here if you wanted to, but I am using plain old white. The keyboard shortcut for Fill is Shift + F5, and here is the result...

After performing the Fill, the image looks like a solid white canvas. But we still have our selection, and we will want to inverse it again by pressing Ctrl + Shift + I. Then press delete to clear/erase the selection...

Press Ctrl + D to deselect and get a clearer view of the image. Now it's time to add a slight Gaussian Blur to make the edges of our "cutout" nice and smooth...

The wonderful PNG image format will allow the blur to accurately blend with anything behind it. To save the image, choose "Save for Web" from the File menu. Be sure to select PNG-24 image format with transparency...

Below is the final result, with a two-pixel black border added to it. As you can see, it blends in with the background perfectly...

Utilizing The Flexibility

It is fairly easy to bring this into a web page and add a hover effect. All you essentially need to do is apply a background color to the image with CSS. I've setup a live demo page which shows a few different ways to implement this idea, including some that use SEO-friendly image replacement.

View the Demo
Download cutout.zip (7kb)

And that's all the time I have today, folks. Please let me know if you found this useful.

Consuming Web Services with DIME Attachments in VB.NET

Josh Stodola - josh420@josh420.com — Mon, 16 Jun 2008 14:38:19 -0600

In this post I will share a solution that I reached the other day regarding an InvalidOperationException that I was receiving when attempting to consume a SOAP web service with DIME attachments. Hopefully this will save you some time, as I was unable to find a good solution online when I encountered the error. DIME attachments are a Microsoft standard that has recently been deemed obsolete. Basically, it's a method of transmitting binary data via SOAP, in which the raw data is delivered after the SOAP envelope. By keeping the data outside of the envelope, it does not have to be Base64 encoded (which would be required to keep the XML valid). Subsequently, the attachment does not have to be consumed by an XML parser on the client-side. This approach dramatically increases the performance of both the client and the server. Although effective, DIME has already been replaced by a newer standard called MTOM. However, DIME is still used today by several established APIs and will probably continue to be used unless an upgrade is absolutely warranted. To work with DIME attachments, Microsoft Web Service Enhancements 2.0 will need to be used.

In my efforts to invoke the web service method and download an attachment, I constantly received the following exception...

InvalidOperationException: Client found response content type of 'application/dime', but expected 'text/xml'

This was somewhat puzzling to me because I did not know what was forcing my client to expect "text/xml". I was hoping there was a ContentType setting I could simply modify that would fix the error, but I was unable to find one. At this point, I was certain that the root of the issue was under the covers. In other words, it wasn't my code causing the problem, it had to be one of the underlying dependencies. So, I went ahead and did some digging into the MSDN documentation and found an article on the topic. One paragraph in particular grasped my attention, specifically the last sentence...

In order to use the WSE DIME support with a Microsoft .NET client application in Visual Studio, you must add a reference for the Microsoft.Web.Services2.dll assembly. Also, after adding a Web reference to the DIME-based Web service, you must modify the proxy class in the References.vb file so that it inherits from the Microsoft.Web.Services2.WebServicesClientProtocol class in WSE.

Sweet! At this point, I thought I had this all figured out. Apparently, the exception I was receiving was caused by the proxy class that is automatically generated by ASP.NET. By default, this class inherits System.Web.Services.Protocols.SoapHttpClientProtocol. According to MSDN, this inheritance will not work with DIME attachments, so it needs to be changed to a class that comes with the WSE toolkit. By using Lutz Roeder's Reflector, which is a wonderful tool, I verified that the only acceptable content types in the default proxy class are text/xml and application/soap+xml...

So that was definitely the problem. This glimmer of hope was shattered when I was completely unable to find the generated proxy class in my solution. Where on Earth is this References.vb file located?! It did not exist in my project because it was a "Web Site" project, and therefore the proxy class was being generated based on the WSDL and compiled on-the-fly by the web server. How do you modify code that is generated on-demand, behind the scenes, by ASP.NET? You don't. Damn!

This frustrated me quite a bit, especially since MSDN did not mention anything about my situation, which is one that I would consider to be common. What I ended up having to do was create a new class library with the sole purpose of invoking the web service. Then I replaced my web reference in the web site project with a reference to this new class libary. This theoretically simple process turned out to be a little confusing, so allow me to step you through it (note that I am using Visual Studio 2005). To begin, I added a new Class Library project to the current web site...

As already mentioned, this class library will solely perform the task of invoking the web service. So, the web reference needs to be removed from the web site project and then re-added to this new class library. When you do this, Visual Studio will generate a proxy class for the service that you will be able to modify accordingly to prevent the InvalidOperationException from being thrown. Here is a screenshot illustrating where the proxy class is located in Solution Explorer...

Open up the Reference.vb file (if you code in C# it will be called Reference.cs) so that you can modify the Inherits statement per MSDN's recommendation...

Namespace WebReferenceName

    Partial Public Class ServiceName
        Inherits Microsoft.Web.Services2.WebServicesClientProtocol
        Inherits System.Web.Services.Protocols.SoapHttpClientProtocol

And that should be the only thing that needs changed. Prior to building, you may want to go into the class library configuration and set it to release mode so that debugging symbols are not generated. Once the library builds successfully, you will need to add a reference to it in your web site project. When you need to call the method that returns DIME attachments in your web site project, simply instantiate the web service within the class library like this...

Dim Service as New MyClassLibrary.WebReferenceName.ServiceName()

This will prevent the InvalidOperationException from being thrown. Every time you update the web reference in the class library, you will have to go back in and change the inheritance of the proxy class again, which is pretty lame. Unfortunately, this is the best solution I was able to come up with. I hope somebody finds this helpful. If you have any questions or relevant insight, please leave a comment.

Support CSS Naked Day Redux

Josh Stodola - josh420@josh420.com — Mon, 07 Apr 2008 21:59:53 -0600

A couple days ago I made a blog post about a simple HTTP handler written in ASP.NET. The purpose? To spit out 404 errors for .css file requests on CSS Naked Day. That handler will undoubtedly work great, but it does have one minor maintenance-related flaw. Each year you will have to update the constant date value to be accurate. Granted, that is a simple task, but it still requires time and a little bit of effort. Dustin Diaz left a comment notifying me of a nice JSON file sitting out on GoogleCode that I could tap into programmatically to determine the official date of CSS Naked Day. And that eliminates the maintenance flaw. Brilliant! So I went ahead and re-wrote the handler to call that JSON file and store the value in cache. There are some constant variables declared at the top that you can modify to your liking. Here is the code…

Imports System.IO
Imports System.Net
Imports System.Web
Imports System.Web.Caching

Public Class Naked
    Implements IHttpHandler

    Private Const NAKED_DAY_API As String = "http://php-naked-day-api.googlecode.com/files/config.json"
    Private Const NAKED_DAY_DEFAULT As Integer = 9 'The default day to use if the API call failed or was invalid (paranoia)
    Private Const CACHE_KEY As String = "Css.NakedDay"
    Private Const CACHE_LIFETIME As Integer = 24 'Number of hours to store the value in Cache

    Public Sub ProcessRequest(ByVal context As System.Web.HttpContext) Implements IHttpHandler.ProcessRequest
        Dim Path As String = context.Request.PhysicalPath

        With context.Response
            If File.Exists(Path) AndAlso Date.Today <> GetNakedDay(context.Cache) Then
                .ContentEncoding = Encoding.UTF8
                .ContentType = "text/css"
                .WriteFile(Path, False)
            Else
                .StatusCode = 404
                .End()
                .Close()
            End If
        End With
    End Sub

    Private Function GetNakedDay(ByRef Cache As Cache) As Date
        Const MONTH As Integer = 4
        Dim NakedDay As Date

        If Cache.Item(CACHE_KEY) Is Nothing Then
            Dim Token As New Regex("""day"" : (\d{1,2})")
            Dim Api As String

            NakedDay = New Date(Date.Now.Year, MONTH, NAKED_DAY_DEFAULT)

            Try
                Using Client As New WebClient()
                    Api = Client.DownloadString(NAKED_DAY_API)
                End Using
            Catch ex As WebException
                Api = String.Empty
            End Try

            If Not String.IsNullOrEmpty(Api) AndAlso Token.IsMatch(Api) Then
                Dim Hold As String() = Token.Matches(Api).Item(0).Value.Split(":")
                NakedDay = New Date(Date.Now.Year, MONTH, CInt(Hold(1).Trim()))

                'Only add the value to cache if the API call was successful
                Cache.Add(CACHE_KEY, NakedDay, Nothing, Date.Now.AddHours(CACHE_LIFETIME), Cache.NoSlidingExpiration, CacheItemPriority.Normal, Nothing)
            End If
        Else
            NakedDay = Date.Parse(Cache.Get(CACHE_KEY))
        End If

        Return NakedDay
    End Function

    Public ReadOnly Property IsReusable() As Boolean Implements IHttpHandler.IsReusable
        Get
            Return False
        End Get
    End Property
End Class

To implement this handler into your ASP.NET web site, simply modify the <httpHandlers> element within <system.web> in the web.config file as illustrated here…

<system.web>
  <httpHandlers>
    <add verb="*" path="*.css" type="Naked" validate="false"/>
  </httpHandlers>
</system.web>

That's it. Enjoy!

Support CSS Naked Day Easily on Your ASP.NET Web Site!

Josh Stodola - josh420@josh420.com — Sat, 05 Apr 2008 19:14:56 -0600

UPDATE: I have came up with a better zero-maintenance solution thanks to a comment by Dustin. The one bellow will still work, you just have to maintain the date constant that is defined at the top of the handler.

I kinda like the idea of CSS Naked Day, which was thought up by Dustin Diaz (a Google fellow) a couple of years ago. I have decided to participate. If you have a personal web site, I would encourage you to do the same. If your web site runs on ASP.NET (highly recommended), you can use the HTTP handler below. I wrote it today and am implementing on my blog. On CSS Naked Day, which is defined at the top of the handler in a constant, it will return a 404 HTTP status code for each and every request for a .css file. You just have to change this date every year and you'll be all set without having to modify any markup or move around files. Your blog engine might already have an HTTP handler setup to intercept .css file requests. If that is true, you can just reference my handler and modify your current one accordingly. Here is the code in VB.NET (attention crazy people: go here to convert to C#)…

Imports System.IO
Imports System.Web

Public Class Naked
    Implements IHttpHandler

    Const NAKED_DAY As Date = #4/9/2008#

    Public ReadOnly Property IsReusable() As Boolean Implements IHttpHandler.IsReusable
        Get
            Return False
        End Get
    End Property

    Public Sub ProcessRequest(ByVal context As System.Web.HttpContext) Implements IHttpHandler.ProcessRequest
        Dim Response As HttpResponse = context.Response
        Dim Path As String = context.Request.PhysicalPath

        If File.Exists(Path) AndAlso Date.Today <> NAKED_DAY Then
            Response.ContentEncoding = Encoding.UTF8
            Response.ContentType = "text/css"
            Response.WriteFile(Path, False)
        Else
            Response.StatusCode = 404
            Response.Close()
        End If
    End Sub
End Class

To implement this handler into your ASP.NET web site, simply modify the <httpHandlers> element within <system.web> as illustrated below…

<system.web>
  <httpHandlers>
    <add verb="*" path="*.css" type="Naked" validate="false"/>
  </httpHandlers>
</system.web>

Talk about easy! ASP.NET freakin' rules. It's uncomparable to Pretty Horrific Programming, I can tell you that much.

Opening External Links in a New Window Based on User Preference

Josh Stodola - josh420@josh420.com — Fri, 04 Apr 2008 18:03:39 -0600

I spend alot of time surfing the Internet. Who doesn't? When I come across an outgoing link (when I say outgoing, I mean a link to a page that resides on a different domain), I always have to make a decision. As an end user, I am clueless as to whether or not the link will open up in a new window. Perhaps the browser is to blame for this. Regardless, I don't usually take any chances, so I right-click the link and choose to open it in a new window/tab. Every potentially suspicious link (most of them are) requires at least two clicks. I have reached the point that I do this instictively, without even thinking about it. I honestly feel the browser should be responsible for telling the user which links will be opened externally, and perhaps there should also be settings in place for the user to specify how they want all links to behave, depending on anchor tag attributes such as target="_blank" and rel="external". It doesn't take long before you realize that waiting for browsers to do what you want is a lost cause, no matter what religion you are or how often you pray. So, I took action. I came up with a Javascript solution that puts the choice in the user's hands.

In my opinion, an ideal solution would be to setup a link/button/checkbox/whatever on your site that, when the user clicks it, will execute some Javascript to force all external links to be opened in a new window. On top of that, the users preference should be "remembered" by storing it in a cookie. You might have seen something like this before. I think this is a pretty damn user-friendly approach. In light of that, I have prepared an elegant script that performs the duty, and is quite simple to implement and extend. Before getting to the juicy code, allow me to explain how it works.

Nuts and Bolts, Baby!

When the web page loads, my script will look for the existence of a cookie. The cookie it's looking for will be one that the same set of Javascript previously created. If it finds it, it will loop through the links accordingly and attach an onclick event handler to open the HREF in a new window (or tab, depending on how your browser settings are configured). If it does not find the cookie, it does nothing. So users without the cookie will have the default link behavior; the target attribute of the links will still be effective. You, the developer of the web site, are responsible for coming up with a stylish link/button/checkbox for the user to make their selection. All you have to do is call a Javascript function when the user clicks this element, with one boolean parameter, and that's it. The rest is done for you. Passing true to the function will make all outgoing links open in a new window. Passing false will make them all open in the same window. This function sets the cookie and loops through every <a> tag on the page. If the HREF attribute points to an external web site, it will be opened in a new window when clicked (if the boolean parameter was true). It's a relatively simple concept, right? Turning concept into code is not as simple. Luckily for you, I've got it all taken care of.

I created an unobtrusive object model that is straight-forward and is nice and simple to use. Most importantly, you can add the feature to an existing site pretty easily. I have tested the code in all major popular browsers as of today (April 2, 2008), and it worked great in all of them. If you have any questions on how to modify this script to your liking, don't hesitate to ask. Here is the Javascript code…

function addLoadEvent(func) {
  if(typeof window.onload != 'function')
    window.onload = func;
  else {
    if(func) {
      var oldLoad = window.onload;

      window.onload = function() {
        if(oldLoad) oldLoad();
        func();
      }
    }
  }
}

var Links = {
  IsCapable: (document.getElementsByTagName),
  CookieId: 'MyLinkCookie',
  GetCookie: function() {
    var vals = document.cookie.split(';');

    for(var i = 0; i < vals.length; i++) {
      var val = vals[i].replace(/^\s+/, '');

      if(val.indexOf(Links.CookieId) == 0)
        return val.substring(Links.CookieId.length + 1, val.length);
    }

    return null;
  },
  SetCookie: function(bool) {
    var dat = new Date();
    dat.setTime(dat.getTime() + 63072000000); //2 years

    var val = Links.CookieId + '=' + bool;
    var exp = '; expires=' + dat.toGMTString();
    var path = '; path=/';

    document.cookie = val + exp + path;
  },
  RemoveCookie: function() {
    document.cookie = Links.CookieId + '=false; expires=Fri, 02-Jan-1970 00:00:00 GMT; path=/';
  },
  SetExternals: function(bool) {
    Links.SetCookie(bool);
    Links.Scan();
  },
  Scan: function() {
    var bool = Links.GetCookie();

    if(!Links.IsCapable || !bool)
      return;

    var links = document.getElementsByTagName('a');

    for(var i = 0; i < links.length; i++) {
      var href = links[i].getAttribute('href');

      if(href.substring(0, 5) == 'http:' || href.substring(0, 6) == 'https:') {
        var pageUrl = location.protocol + '//' + location.host;

        if(href.substring(0, pageUrl.length) != pageUrl) {
          if(bool == 'true') {
            links[i].onclick = function() {
              var winId = 'win' + Math.random().toString().substring(2);
              var newWin = window.open(this.href, winId);
              newWin.focus();
              return false;
            }
          }
          else
            links[i].onclick = null;

          links[i].removeAttribute('target');
        }
      }
    }    
  }
}

addLoadEvent(Links.Scan);

Uhhhh, How Do I Use This Code?

I won't leave you hangin' with just a chunk of Javascript; I have setup a demo page illustrating the functionality. It is simple, really. All you have to do is bring in the script into your page (with <script> tags, respectively). There are two main functions: Links.SetExternals and Links.RemoveCookie. SetExternals is the function I mentioned earlier that expects the boolean parameter. Calling the RemoveCookie function will essentially return link behavior to the default.

The demo uses standard hyperlinks to set the preference, but I think a checkbox is probably the most practical thing to use here, which can be done rather easily just like this...

<input type="checkbox" onclick="Links.SetExternals(this.checked);" />

So go ahead, make your visitors happy by giving them a useful option. If you have any problems using this script, or any relevant questions, feel free to express yourself. That's all folks, enjoy!

Scott Guthrie Used to be an Actor!

Josh Stodola - josh420@josh420.com — Sat, 08 Mar 2008 20:13:25 -0600

So… I was watching a classic film called Bio-Dome today, and I couldn't help but notice that an actor in that film looks exactly like Scott Guthrie! I was expecting this feller to start yip-yapping about the glorious benefits of ASP.NET, but then I realized that I was not watching a podcast…

That screenshot is at about 13:38 into the movie. Am I crazy, or does this guy look exactly like ScottGu?! Alright, so the real Scott isn't that bald, but still. This guy could show up at a developers conference and get numerous hand-shakes from people he doesn't know.

For the instinctively curious, this actor's real name is Kevin West.

Twitter Sucks!

Josh Stodola - josh420@josh420.com — Thu, 28 Feb 2008 22:59:34 -0600

I decided to sign up for Twitter the other night for no reason at all. I have used it a few times since then, and I can appreciate the whole social networking aspect of it. However, I can't help but conclude that this the most idiotic thing that I have ever signed up for. I just don't understand the purpose of writing several short messages throughout the day. Are you going to return several months later to see what you did on Feburary 28th? Oh, and it's totally unstable and slower than molasses! After Googling about it, I noticed on Wikipedia that it is written in Ruby on Rails. LMAO! That explains why the site is down RIGHT NOW!

Another thing worth mentioning is the poor site design. It literally took me four or five clicks just to locate a textbox that I can enter a name to search. When I naturally click "Find & Follow", this is the crap I am greeted with…

I don't have Hotmail or Yahoo or Gmail or AOL or MSN. Even if I did, I probably would not be giving up the password. So I have one other option: send an email invitation. WTF?! How about a textbox to enter a name, and when I press a button, you look in your database of users for a match and display the results to me as soon as possible. Wow, that would be amazingly brilliant for a page with a title of "Find People You Know on Twitter".

I am sure this service gets most of its usage via mobile phones, 3rd party applications and what-not, but c'mon, the web interface should be usable as well. After all, is that not the "roots" of the application?

I gotta say that Twitter and Gravatar give me horrible impressions of Ruby on Rails. I think the word I am looking for is scalability.

The Underlying Connection was Closed - How to Fix This Pesky ASP.NET WebException

Josh Stodola - josh420@josh420.com — Sat, 16 Feb 2008 00:51:52 -0600

If your ASP.NET application calls a web service (you have a reference in your App_WebReferences folder), you might come across this fairly typical System.Net.WebException. I have seen two different breeds of this exception, and the solutions for each of them are quite similiar. I will start off with the most common of the two.

The underlying connection was closed: A connection that was expected to be kept alive was closed by the server.

This is the one you are most likely to see, and the fix for this is simple and generally well-known. You need to create a class that inherits your web service class. Then you simply override the GetWebRequest function and modify the KeepAlive HTTP header accordingly…

Public Class MyWebService
    Inherits ServiceName.WebService

    Protected Overrides Function GetWebRequest(ByVal uri As Uri) As System.Net.WebRequest
        Dim webRequest As System.Net.HttpWebRequest = MyBase.GetWebRequest(uri)

        webRequest.KeepAlive = False

        Return webRequest
    End Function
End Class

Simple as that. When you need to invoke the web service elsewhere in your code, just instatiate that class instead of what you would do normally…

Dim Response as String

Using MyService as New MyWebService()
    Response = MyService.METHOD_NAME("MethodInput")
End Using

And that takes care of that exception.

The underlying connection was closed: The connection was closed unexpectedly.

Sometimes, using the solution above will cause this exception. I believe that it has something to do with the way Windows Authentication is configured in IIS. I ran across a situation where I had to deploy my applicaton (with web service references) to a remote server that was not in our control; it was being hosted for us. Apparently, they were using NTLM Authentication, which requires connections to be kept-alive, hence this exception. However, using KeepAlive makes your requests totally inconsistent, in most cases. They will work one time, but will fail the next. The host was not willing to change for this for our application, obviously, so I was stuck. Finally, I was able to determine that the ConnectionGroupName property is what is used to re-establish the connection that was supposed to be "kept alive". So, you can just set that property to a random value (GUID), in the same way that I set the KeepAlive property above…

Public Class MyWebService
    Inherits ServiceName.WebService

    Protected Overrides Function GetWebRequest(ByVal uri As Uri) As System.Net.WebRequest
        Dim webRequest As System.Net.HttpWebRequest = MyBase.GetWebRequest(uri)

        webRequest.ConnectionGroupName = Guid.NewGuid().ToString()

        Return webRequest
    End Function
End Class

Problem solved; that worked fantastic! I hope this has helped you out, and I also hope that I have had my final bout with the System.Net.WebException.

Creating a "Save As..." Button for the Rendered Markup of an ASP.NET Page

Josh Stodola - josh420@josh420.com — Sat, 02 Feb 2008 16:51:11 -0600

I am not sure why you would want to do this, but I have seen the question asked on the ASP.NET forums a couple of times so I figured I would post a solution for it. It is not that difficult to create a "Save As…" button on your web page that, when clicked, will prompt the user with a download dialog containing the rendered markup of the page. All you have to do is have a button that will refresh the page, except append a query string to it. Then in the Page_Load you can check for the existence of the query string and change the content type and disposition of the response to trigger the dialog.

First, add the following markup where you want it on your page…

<input type="button" value="Save Page As…" onclick="(location.href.indexOf('?') == -1) ? location.href += '?save=true' : location.href += '&save=true';" />

All that is doing is adding a query string to the existing URL and redirecting to it. Add this code in the Page_Load to check for the query string and do the necessary steps to force the download prompt…

If Request.QueryString.Item("save") IsNot Nothing Then
    If Request.QueryString.Item("save").ToString() = "true" Then
        Response.Clear()
        Response.ContentEncoding = Encoding.UTF8
        Response.ContentType = "text/plain"
        Response.AddHeader("content-disposition", "attachment")
    End If
End If

And that is all there is to it! This could easily be wrapped up in an ASP.NET User Control so that it could be dropped into several different pages effortlessly. Again, I am not sure why you would want to do this. To me, this is the same as manually viewing the page source and saving it to your hard drive. Oh well, I hope somebody out there finds it useful.

UPDATED 02/03/2008: Shortened onclick event handler for the button.