Friday, January 29, 2010

ASP.NET Controls, even the 3rd party controls are great, until you really try to use them

Over the years I have had a kind of love hate relationship with asp.net web controls and after having some frustration recently I decided to rant about it.

Conceptually it’s nice idea, the asp.net controls abstract away a lot of the “web development” so you can focus on just creating your application and not futzing around with html inputs, and creating elaborate javascript, and they give you a kind of sort of state;  and for about 90% of the time they do everything you want them to do. 

Were they fall down is really in two places:

  1. In all reality not getting that last 10% really isn’t the controls fault it’s trying to be everything to every body and you really can’t do that.  This is even true the for expensive 3rd party controls, granted vendors like DevExpress, Component One, and Telerik make some very nice controls (a lot better then the stock controls from MS) I would even say they give you an additional 5%, and for the most part they make developing web forms easier and faster, BUT they still fall down when you really push them hard and then it feels like your spending all of your time tweaking their control verses just getting your application written. 
  2. The cost of getting that statefullness by using viewstate can sometimes crush your application, like most things it’s fine in moderation, unfortunately ViewState can quickly explode in size and blot your application, creating HUGE page sizes, slow responses and if it gets big enough it starts to corrupt and your applications starts to error, fun! Once again the 3rd party vendors tend to make better controls with less viewstate.

So what is the solution, first learn about webforms controls and how they work, then create some server controls of your own.  If you look at previous posts I talk about how to create different types of controls and I’m only scratching the surface.  By rolling your own controls you can focus on getting them to do exactly what you want them to do and that might be the solution to getting the last 5% of the functionality you need.  Using Asp.net controls is one of the places where it’s really easy to get started with very little knowledge and quickly get sucked under.  This is defiantly a place where if you don’t understand how it works it can and will come back to get you. 

Finally if you can’t get that last little bit of functionality, well for the most part you just have to deal with it.  Sorry to say but we are getting to a point where the development work on the web is going beyond what the underlying technology just can support. 

Monday, January 18, 2010

Unit Testing For Javascript

Recently I had to write some fairly substantial javascript, and in the process I was reminded of how fast javascript, like most scripting languages, and get out of hand really quick. I took a step back and thought why should I work in javascript any different then I do in other language? Just because it’s a scripting language doesn't make it any less valid of a development platform, and I can apply the same design pattern and principles I would with anything else.

So The first thing I did was went looking for a unit testing framework and I found JSUnit. JSUnit is available from http://www.jsunit.net/ and is a port of the JUnit framework to javascript. The entire testing framework and test runner is written in javascript so I can use it anywhere you use javascript and in any development platform(php, ruby, Perl, java, .net, etc).

Getting started I have this javascript function I want to test

   1: function addTwoNumber(x, y) {
   2:     var result = x + y;
   3:     return result;
   4: }

so first I create an html page and add the jsUnitCore.js and my .js file containing this function (almost all of your javascript really should be in .js files for a number of reasons, but that’s another blog post).

   1: <html>
   2:     <head>
   3:         <script language="JavaScript" src="jsUnitCore.js"></script>
   4:         <script language="JavaScript" src="javascriptCode.js"></script>
   5:     </head>
   6:     <body>        
   7:     </body>
   8: </html>

then to add test cases I create functions inside a script tag in the body of the html page like so

   1: <body>
   2:     <script type="text/javascript">
   3:         function test_addTwoNumber_TwoNumbersAreAdded() {
   4:             var result = addTwoNumber(2, 2);
   5:             assertEquals("Adds 2 values together", 4, result);
   6:         }
   7:  
   8:     </script>
   9: </body>
to tag a function as a test, simply start the function off with the text “test” and the test runner will pick it up and run it. The assert is done with the function assertEquals provided by the jsUnitCore.js file and takes 3 prams (test description, expected value, and actual result) and to run the test you simply open the testRunner.html page in the jsunit dir in a browser

image

then add the file location and run the tests, it’s that simple. If you look at the screen shot you will notice that I’m using opera, for what ever reason I couldn’t get the test runner to work in Firefox, every time I would try and run my test file I would get a time out error, The funny thing is it works in IE, Chrome, and Opera just fine, weird.

Saturday, December 12, 2009

Using WCF to build fast and lean web applications

Recently I ran into a problem where I needed to reduce the size of a page, we had a page with a lot of drop down controls that may or my not be used. The problem with this was the size of the html was huge, every one of the controls was fully populated with over 50 items, next the view state was equally huge storing all of that content. The Solution, using WCF to load the control’s content on demand.

Starting off we need to get WCF configured, and this confused me, I had to manually register WCF in IIS, WHAT!! ok to save you the hassle of having to research how to do this, execute c:\WINDOWS\Microsoft.NET\Framework\v3.0\Windows Communication Foundation\ServiceModelReg.exe –i and your good to go.

We have two options for populating the dropdowns, you can use a 3rd party control like Telerik’s radcombobox, or roll your own, which in some ways is fairly simple but rather involved, for right now I’m going to show you how to roll your own, to see how to use radComboBox see here.

Lets start with the WCF service, create a Ajax enabled WCF service. For what ever reason I had to remove the namespace to get this to work, and the examples I found online had it removed as well, this is something I’m going to research more on later. Next we need to build a data entity to hold the data, nothing fancy just something like this

   1: public class Item
   2: {
   3:     public string Text { get; set; }
   4:     public string Value { get; set;}
   5: }
   6:  
   7: public class Items : Collection<Item>
   8: {}
next we build the service, in this case I’m just building some data.
   1: [ServiceContract(Namespace = "")]
   2: [AspNetCompatibilityRequirements(RequirementsMode = AspNetCompatibilityRequirementsMode.Allowed)]
   3: public class DataService
   4: {
   5:     [OperationContract]
   6:     public Items GetData()
   7:     {
   8:         Items myItems = new Items();
   9:         for (int i = 0; i < 5; i++)
  10:         {
  11:             Item newItem = new Item()
  12:                {
  13:                    Text = string.Format("value {0}", i), 
  14:                    Value = i.ToString()
  15:                };
  16:             myItems.Add(newItem);
  17:         }
  18:         return myItems;
  19:     }
  20: }
when you add the Ajax enabled WCF service some items will be added to the system.serviceModel section of the web config in this case it looks like this
   1: <system.serviceModel>
   2:   <behaviors>
   3:    <endpointBehaviors>
   4:     <behavior name="DataServiceAspNetAjaxBehavior">
   5:      <enableWebScript />
   6:     </behavior>
   7:    </endpointBehaviors>
   8:   </behaviors>
   9:   <serviceHostingEnvironment aspNetCompatibilityEnabled="true" />
  10:   <services>
  11:    <service name="DataService">
  12:     <endpoint address="" behaviorConfiguration="DataServiceAspNetAjaxBehavior"
  13:      binding="webHttpBinding" contract="DataService" />
  14:    </service>   
  15:   </services>
  16:  </system.serviceModel>
if your not able to get your service to work with the Namespace, and you remove it, make sure to also remove it from the contract attribute in the endpoint node of the service node.

Next we add the service reference to the page using the script manager tag
   1: <asp:ScriptManager ID="scriptManager" runat="server">
   2:     <Services>
   3:         <asp:ServiceReference Path="/DataService.svc" />
   4:     </Services>
   5: </asp:ScriptManager>
Now we can access the webservice with javascript. First we need to have a dropdown control to work with, for this we are going to just use a plain old html select tag and add an ID and a runat=”server”.
   1: <select 
   2:     id="dropDownList" 
   3:     runat="server" 
   4:     enableviewstate="false" 
   5:     style="width:200px;" 
   6:     onfocus="populateDropDown(this);">
   7: </select>    
something else I have done here is I set enableviewstate="false" this solution doesn't use any view state so for the “cool kids” out there using MVC, this solution works for you as well. Next add the javascript to populate our dropdown, as you can see in the select tag we have a populateDropDown function that is called with onfocus.
   1: var Control;
   2: function populateDropDown(dropdown) {
   3:     Control = dropdown;
   4:     DataService.GetData(dropDownRequest_onSuccess, onFailure);
   5: }
Accessing the service is made easy by using the script manager, all we have to do is call javascript object created by the script manager with the same name as the service contract in this case it’s “DataService” and call the function with the same name as our service method “GetData”. The function takes 2 delegate functions that act as event handlers as parameters, in this case it’s dropDownRequest_onSuccess and onFailure. At this point I would like to say I’m in no way a javascript expert, if you know of a better way to any part of the javascript code by all means do it, and let me know.
   1: var selectedOptionValue;
   2:  
   3: function dropDownRequest_onSuccess(result) {
   4:     Control.options.length = 0;
   5:     for (var x = 0; x < result.length; x++) {
   6:         var opt = document.createElement("option");
   7:         opt.text = result[x].Text;
   8:         opt.value = result[x].Value;
   9:         if (result[x].Value == selectedOptionValue) {
  10:             opt.setAttribute("selected", "true");
  11:         }
  12:         Control.options.add(opt);
  13:     }
  14: }
  15:  
  16: function onFailure(result) {
  17:     window.alert(result);
  18: }
The dropDownRequest_onSuccess function clears (if any) options from the select tag, then loops though the result from the webservice, creating option elements, setting their text and value then checking to see if the value matches the value of the selectedOptionValue var and if they do it sets that option as selected then adds it to the options collection of the select control.

The onFailure is called when the service call fails, and for this we are just doing a simple alert to say it failed.

The next part is getting the data out when the page is submitted, because we are not using any view state we can’t get the value the same way we would with a regular asp.net control, we have to pull it from Request.Form collection.
   1: string value = Request.Form[dropDownList.ID];
Finally we need to add a way to set an option as selected. The code to set the selected option is already there we just have to utilize it by setting the value of var selectedOptionValue, to do this we use the ClientScript.RegisterStartupScript like this
   1: private void setValue()
   2: {
   3:     string script = string.Format("setpopulateDropdown('{0}',{1});", dropDownList.ClientID, Request.Form[dropDownList.ID]);
   4:     ClientScript.RegisterStartupScript(typeof(Page),"setDropdownValue",script,true);
   5:     
   6: }

to call this javascript function

   1: function setpopulateDropdown(controlId, value) {
   2:     Control = document.getElementById(controlId);
   3:     selectedOptionValue = value;
   4:     populateDropDown(Control);
   5: }

This sets the control to be updated, set the selected value, then populates the the control from the webservice.

Like I said it’s kind of a long process but you can cut down your page size a lot if your doing multiple dropdowns with large amounts of data that may not be used.

As always the full source code for this project is available (here).

Thursday, October 22, 2009

First look at Visual Studio 2010 Beta 2

Visual studio 2010 Beta 2 is out, so lets take a look at some of the changes:

  1. The new WPF UI is a nice clean face lift.
  2. The new extension manager makes getting, updating, and managing plug-ins, templates, and controls(Telerik has almost their entire product line available) from the web easy, my only real issue with the extension manager is I can’t find a way to load plug-ins that aren’t in the gallery, and a lot of my favorite plug-ins(Ghost doc and R#) aren’t supported yet.
  3. The removal of the Team Studio versions, the Team Studio versions where confusing and over priced. The new SKUs are VS 2010 Professional, VS 2010 Premium, and VS 2010 Ultimate. For the full product descriptions see the Visual Studio Product page.

Some of the things that didn’t change:

  1. Adding a reference is still slow and painful.
  2. The testing frame work really hasn't changed much, although the test seem to run faster.

Things that are not quite there yet:

  1. Intellisense doesn’t work very well, but this may be a configuration setting.
  2. The number of supported plug-ins is limited, but that is understandable seeing that the plug-in model has completely changed

All in all I really like Visual Studio 2010 and I’m really excited for when it’s ready for prime time.

Saturday, October 17, 2009

Taking the pain out of code documentation

I have never known a developer that liked writing document let alone writing code document, well I take that back there was one guy but he had *Issues* and we’ll just leave it at that. I have heard all of the excuses, and in the end it still comes down to, we would rather write code then write about the code.

One common statement coming from the TDD/DDD methodologies is that the unit tests document the code; while this is great if we have the unit test, but even then a little documentation would still be useful so we don’t have to dig though unit test to figure out how to use a library. Another favorite excuse is the method names document themselves, granted having intelligently named methods and variables make understanding the code a lot easier(and we’ll get into that latter) but short of having the average method name be 40+ chars long this really isn’t practical either.

Luckily the are some tools to make writing code documentation much easier. For starters the .NET compiler can be configured to generate XML documentation files by setting an xml output file in the project build configuration. The xml content is created from specially formatted XML comments that looks like this

   1: /// <summary>
   2: /// Saves the contact.
   3: /// </summary>
   4: /// <param name="contact">The contact to be saved.</param>
   5: /// <returns>
   6: /// A true or false result of the successful saving this contact
   7: /// </returns>
   8: public bool SaveContact(Contact contact)

You can write this by hand or use a tool like the GhostDoc plug-in for Visual Studio. GhostDoc will auto generate documentation based on configurable rules for the different objects, classes, method, properties, etc. One of the very nice features about being able to configure the documentation rules is you can add additional item like remarks that say when it was documented and by who.

The second nice thing about using GhostDoc is the rules used to figure out what to put into the documentation can be used to make sure your using intelligent names, it’s not always right but it can be a good indicator for how good of names your using.

So why add this? This gives you two advantages, first it adds intelligence for your code, next it allows you to create MSDN like documentation using the .NET documentation generator Sandcastle.

Sandcastle can be downloaded from http://sandcastle.codeplex.com/ and provides a command-line interface for building your documentation from the .dll and .XML file generated by the .net compiler making it easy to add documentation generation to your automated build system. For the user it’s a little clunky and painful to use, if you want to build the documentation without scripting, there is an app for that, the Sandcastle Help File Builder located at http://www.codeplex.com/SHFB.

sandcastle_builder

Basically this lets you build a configuration that can include what is documented, the layout of the documentation, adding headers, footers, errors for missing items etc. and when you’re all said and done you get documentation that looks like this

sandcastle_documentation

it’s that easy.