About a month and a half ago I announced that I am writing a book, I was really overwhelmed by the amount of support that I received from this announcement.  Both myself and Al are really looking forward to the day when this book ships, and we start receiving real feedback on all our hard work.  However, both of us would like to start receiving feedback as soon as possible, so…

In an effort to write the book and keep blogging, I decided to write/blog the last chapter, Chapter 2.  I am doing this so I can receive feedback on this chapter as early as possible.  Because this chapter, in my opinion, is probably the most critical of the book, it defines the context around ASP.NET MVC and how it differs from ASP.NET Web Forms, as well as giving a historical perspective of the MVC pattern.

In the next several posts we will cover the following parts of Chapter 2 from the book:

• The Model-View-Controller Pattern
• ASP.NET MVC vs. ASP.NET Web Forms
• Installing the Prerequisites
• Your First ASP.NET MVC Project
• The Model
• The View
• The Controller

ASP.NET MVC 1.0 Website Programming: Problem – Design – Solution by Nick Berardi New: $31.49 This item has not yet been released. You may order it now and we will ship it to you when it arrives.

The Model-View-Controller Pattern

The Model-View-Controller architectural pattern has been around since 1978 and was first described by Trygve Reenskaug while working on a programming language called Smalltalk at Xerox PARC.  The implementation was first described in his now famous paper on the subject, titled Applications Programming in Smalltalk-80: How to use Model-View-Controller, published on December 1979, and has been popping its head up in many different ways and forms since the original paper was published.  Reenskaug maintains a page that explains MVC in his own words (http://heim.ifi.uio.no/~trygver/themes/mvc/mvc-index.html), and contains his publications on the subject; it is well worth the read and is only two pages long.

The MVC pattern has been implemented in most every programming language that is in use today, including ColdFusion, Java, JavaScript, Perl, PHP, Python, Ruby, Smalltalk, XML, and of course .NET.  In fact in November, 2002 the W3C, the main international standards body for the World Wide Web, voted to make the MVC pattern part of their XForms specification, which will be integrated directly into the XHTML 2.0 standard.

Reenskaug explains on this site that “The essential purpose of MVC is to bridge the gap between the human user’s mental model and the digital model that exists in the computer.”  As illustrated in Figure 2-1.

Figure 2-1

He goes on to explain that “The ideal MVC solution supports the user illusion of seeing and manipulating the domain information directly.  The structure is useful if the user needs to see the same model element simultaneously in different contexts and/or from different viewpoints.”  This is important because it puts the emphasis not on the application, but how the user perceives the data, the controller and view is only a means to the end of allowing the user to visualize the model in other words.

Reenskaug defines the Model-View-Controller in the following way.

• Model: Represents knowledge.  A model can be in the simplest case a single object in your application, or in a complex case combination of objects.  It should represent the world as seen by the developer for the application that is being developed, in other words your database or domain.
• View: Visual representation of the Model.  It should highlight the certain aspects of the model while minimizing the others where possible.  According to Reenskaug it should act as a presentation filter.  What he describes as a presentation filter is the notation of a contract created between the Model and the View that will provide the parts of the model requested for the presentation by the View.
• Controller: A controller provides a link between the user and the system.  It provides the user with actions that can be taken against the Model, which in other words creates a set of inputs that can be acted upon and represented to the user in one or more ways through a View.

Bringing MVC Down To Earth

The concepts and ideas behind MVC were honestly a little abstract for me when I was first getting started, it took me a while to understand how the Model, View, and Controller where suppose to work together to create an application.  Unfortunately at the time I didn’t have a great example that clearly defined the lines between the different parts of the Model, View, and Controller, so I had to learn the hard way.  Lucky for us Jeff Atwood, of codinghorror.com fame, provided an example that really struck a chord with me.  Figure 2-2 is a visual representation of his example.

Figure 2-2

This example almost perfectly represents MVC in a way that any web developer with only basic knowledge of HTML and CSS can understand.

• Model: The HTML is the “skeleton” or definition of the data to be displayed to the user.
• View: The CSS is the “skin” that gives the HTML a visual presentation.  The CSS can be swapped out to view the original content in a different manor, without altering the underlying Model.  They are relatively, but not completely, independent of each other.
• Controller: The browser is responsible for combining the CSS and HTML, into a final representation that is rendered out to the screen in the form of pixels.  It gathers input from users, but it is restricted to the input defined by the HTML in the form of input, select, textarea, and button DOM objects.

I find this to be an awesome acknowledgement to the success of the Model-View-Controller, because the browser is a natural interface for a computer user that wants to visualize the World Wide Web.  It successfully maps the Mental Model, from Figure 2-1, that a designer envisioned as an interface for the user to the Computer Model, which a developer coded for use on the World Wide Web.  So I hope this helped you visualize MVC in a way that helps you break out and understand the concepts behind the Model, View, and Controller.  If you would like to read Jeff’s full article it is available at http://www.codinghorror.com/blog/archives/001112.html.

For the purpose of this book we are going to define MVC as the following:

• Model: The classes which are used to store and manipulate the state of the database, through our domain objects, combined with some business logic.
• View: The user interface parts, coded in HTML, necessary for rendering the Model to the user.  It may also render the Model as XML or JSON if needed programmatically by JavaScript.
• Controller: The application layer that will accept the input and save that information to the database through our Model.  It will also contain a small amount of business logic necessary for controlling and validating the inputs.  The controller will also decide which view to render, the HTML, XML, or JSON depending on the form that was requested by the browser.

The above definition of MVC for our application, The Beer House, is an almost exact representation of MVC as defined by the ASP.NET MVC team.

This post is licensed under a different license than the rest of my site. Copyright © Wiley Publishing Inc 2009

Wow that is a long title. Recently I have been looking for quick posts that I can put out each day to keep my blog relevant and also so I don’t feel like I am slacking off too much. Today I want to post about a little known feature in my .NET URL Rewriter and Reverse Proxy (aka. Managed Fusion URL Rewriter) that I have developed in my spare time, mostly out of necessity for this blog and other projects I have worked on.  Here is a quick run through of what it does.

Managed Fusion URL Rewriter is a powerful URL manipulation engine based on the Apache mod_rewrite extension. It is designed, from the ground up to bring all the features of Apache mod_rewrite to IIS 6.0 and IIS 7.0. Managed Fusion Url Rewriter works with ASP.NET on Microsoft’s Internet Information Server (IIS) 6.0 and Mono XPS Server and is fully supported, for all languages, in IIS 7.0, including ASP.NET and PHP. Managed Fusion Url Rewriter gives you the freedom to go beyond the standard URL schemes and develop your own scheme.

But one feature that I added that is not part of the official Apache mod_rewrite documentation is the ability to add custom modules to extend the use of the URL rewriter in non-traditional ways.  One great example of this was born out of wanting to clean up the SEO mess I created in the early days of this blog.  I had to support the following different types of URL patterns:

1. http://www.coderjournal.com/?p=23
2. http://www.coderjournal.com/2008/03/14/some-post.html
3. http://www.coderjournal.com/2008/03/14/some-post

to transform them in to the URL pattern that I finally settled on today:

• http://www.coderjournal.com/2008/03/some-post

In the above list #2 and #3 were pretty easy to transform using the following rules:

RewriteRule ^(/[0-9]{4}/.*).html$    $1/ [NC,R=301]
RewriteRule ^(/[0-9]{4}/[0-9]{1,2}/)[0-9]{1,2}/(.*)$    $1$2 [R=301]

Because they contained all of the elements that make up my current URL.  As you can imagine problems arose when I had to support links that used #1′s syntax.  It contains zero elements that I can use to create my current URL.  Being a programmer who beleives that each part of a system should handle gracefully the domain it was designed to support, in this case a URL rewriter should be able to handle any senario that has to do with URL rewriting.  I added in support that allowed developers to naturally extend the URL rewriter to accomplish any type of URL rewriting task they could think of.

Setting Up the URL Rewriter Rules

In my case I needed to handle the following SQL query everytime I saw a URL that matched #1.

select concat('http://www.coderjournal.com/',year(post_date),'/',month(post_date),'/',post_name,'/') from wp_posts where ID = $1;

What this query does is query the WordPress database table that contains all the posts by the post ID and have it return the actual absolute path to the post, that should be displayed in the URL.  To do this I created a new directive for the mod_rewrite syntax called RewriteModule.  I also had to extend the RewriteRule and RewriteCond directives to support these new module extensions.  The RewriteModule, RewriteRule, and RewriteCond are defined by the following syntax:

RewriteModule <Reference Name> <Namespace>,<Assembly>
RewriteRule[([<Left Module>],[<Right Module>])] <Pattern> <Substitution>
RewriteCond[([<Left Module>],[<Right Module>])] <Test String> <Condition Pattern>

The parts in light blue parts above are optional to creating the rule.  In my case for this blog the rewriter directives looked like the following:

RewriteModule PostQueryString CoderJournal.Rewriter.Rules.PostQueryStringRuleAction, CoderJournal.Rewriter.Rules
RewriteRule(,PostQueryString)   ^/\?p=([0-9]+)$    "select guid from wp_posts where ID = $1;" [R=301]

I have highlighted in red the important parts of the syntax that indicate the custom module processor that should be used on the RewriteRule directive and how it relates back to the class defined in the RewriteModule. 

Creating the Module

I have to warn you that I am not going to demonstrate and show all the properties and methods on the interface that are important for creating a custom module, but I am going to show you the actual meat of the module that is involved in the lookup of the URL from the database.

public Uri Execute(int logLevel, string logCategory, HttpContext context, 
                   Pattern pattern, Uri url, string[] conditionValues, 
                   IDictionary<string, string> flags)
{
	string inputUrl = url.GetComponents(UriComponents.PathAndQuery, UriFormat.UriEscaped);
	string sqlCommand = pattern.Replace(inputUrl, Text, conditionValues);
	string substituedUrl = String.Empty;

	using (MySqlConnection connection = new MySqlConnection(Properties.Settings.Default.DatabaseConnection)) {
		using (MySqlCommand command = connection.CreateCommand()) {
			command.CommandText = sqlCommand;
			command.CommandType = CommandType.Text;

			try {
				connection.Open();
				substituedUrl = command.ExecuteScalar() as string;
			} finally {
				connection.Close();
			}
		}
	}

	return new Uri(url, substituedUrl);
}

It may not be clear right away what is going on, but on line 6, I am replacing the defined value in the regular expression (^/\?p=([0-9]+)$) with the SQL query (from above) to produce a query that will be run against the database. So if the following URL came in to my server:

• http://www.coderjournal.com/?p=372

It would produce a SQL query that looked like this:

select concat('http://www.coderjournal.com/',year(post_date),'/',month(post_date),'/',post_name,'/') from wp_posts where ID = 372;

Notice that the ID, 372, shows up in both the URL and the query, that is because this is the part I am most interested in, in the URL, because it is the only part of the URL that I need to query the database to find the actual path of the post.

Now that we have the query we can execute it on the database, using lines 9 through 21, and create the resulting URL on line 23. The resulting URL is then passed back through the URL rewriter, and processed using the flags defined. In my case [R=301], actually indicates that I want to do a 301 Permanent Redirect on the URL, which tells the browser and search engines, a like, that they need to update their URL for this page.

You can test out the above conditions by using the following URL’s that all redirect back to this page:

1. http://www.coderjournal.com/?p=372
2. http://www.coderjournal.com/2008/12/9/creating-extension-module-net-url-rewriter-reverse-proxy.html
3. http://www.coderjournal.com/2008/12/9/creating-extension-module-net-url-rewriter-reverse-proxy/

The code as always is available on my SVN server at Google Code.

• The Module for Coder Journal
• The Rules File for Coder Journal

I hope this comes in handy to some of you developers that have to support legacy URL’s in your own product or a project that you are working on. As always if you have any questions or need anything clarified please feel free to contact me or leave a comment below.

About a month ago I was experimenting with different ways to optimize my LINQ queries against the IdeaPipe database, in order to improve the read times. I wanted to improve the read times because our new Facebook Application was being launched and I anticipated an increase in our traffic to the server, which is used to host IdeaPipe and the Facebook Application component.

Whenever I am trying to optimize SQL queries I fire up SQL Server Profiler and take a look at how the queries are performing. This helps me identify queries that are taking a longer time to execute and probably need to be looked at or re-thought. One of the queries that I identified as needing improvement was the following LINQ query:

Old Query

from i in source
join xgl in GroupIdeaLinks on i.IdeaId equals xgl.IdeaId into groupLinksGroup
from gl in groupLinksGroup.DefaultIfEmpty()
let visibility = (gl == null ? 'O' : gl.Group.VisibilityPermission)
let groupId = (gl == null ? -1 : gl.GroupId)
where visibility == 'O' || GroupMembers.Count(m => m.GroupId == groupId && m.IsApproved && m.UserId == userId && (visibility == 'G' || (m.RoleId & (int)Role.Manager) == 0)) == 1
select i;

That produced this monster of a SQL statement:

SELECT [t5].[IdeaId], [t5].[CategoryId], [t5].[UserId], [t5].[IsPopular], [t5].[PopularOn], [t5].[Title], [t5].[SafeDescription], [t5].[Description], [t5].[Path], [t5].[Score], [t5].[Rank], [t5].[ExternalLink], [t5].[VideoLink], [t5].[BumpUpCount], [t5].[BumpDownCount], [t5].[TotalBumpCount], [t5].[TotalCommentCount], [t5].[CreatedOn], [t5].[rowversion]
FROM (
    SELECT [t4].[IdeaId], [t4].[CategoryId], [t4].[UserId], [t4].[IsPopular], [t4].[PopularOn], [t4].[Title], [t4].[SafeDescription], [t4].[Description], [t4].[Path], [t4].[Score], [t4].[Rank], [t4].[ExternalLink], [t4].[VideoLink], [t4].[BumpUpCount], [t4].[BumpDownCount], [t4].[TotalBumpCount], [t4].[TotalCommentCount], [t4].[CreatedOn], [t4].[rowversion], [t4].[value],
        (CASE
            WHEN [t4].[test] IS NULL THEN @p1
            ELSE [t4].[GroupId]
         END) AS [value2]
    FROM (
        SELECT [t0].[IdeaId], [t0].[CategoryId], [t0].[UserId], [t0].[IsPopular], [t0].[PopularOn], [t0].[Title], [t0].[SafeDescription], [t0].[Description], [t0].[Path], [t0].[Score], [t0].[Rank], [t0].[ExternalLink], [t0].[VideoLink], [t0].[BumpUpCount], [t0].[BumpDownCount], [t0].[TotalBumpCount], [t0].[TotalCommentCount], [t0].[CreatedOn], [t0].[rowversion], [t2].[test], [t2].[GroupId],
            (CASE
                WHEN [t2].[test] IS NULL THEN @p0
                ELSE CONVERT(NChar(1),[t3].[VisibilityPermission])
             END) AS [value]
        FROM [Ideas].[Ideas] AS [t0]
        LEFT OUTER JOIN (
            SELECT 1 AS [test], [t1].[GroupId], [t1].[IdeaId]
            FROM [Groups].[GroupIdeaLink] AS [t1]
            ) AS [t2] ON [t0].[IdeaId] = [t2].[IdeaId]
        INNER JOIN [Groups].[Groups] AS [t3] ON [t3].[GroupId] = [t2].[GroupId]
        ) AS [t4]
    ) AS [t5]
WHERE (UNICODE([t5].[value]) = @p2) OR (((
    SELECT COUNT(*)
    FROM (
        SELECT
            (CASE
                WHEN ([t6].[GroupId] = [t5].[value2]) AND ([t6].[IsApproved] = 1) AND (([t6].[UserId]) = @p3) AND ((UNICODE([t5].[value]) = @p4) OR (([t6].[RoleId] & @p5) = @p6)) THEN 1
                WHEN NOT (([t6].[GroupId] = [t5].[value2]) AND ([t6].[IsApproved] = 1) AND (([t6].[UserId]) = @p3) AND ((UNICODE([t5].[value]) = @p4) OR (([t6].[RoleId] & @p5) = @p6))) THEN 0
                ELSE NULL
             END) AS [value]
        FROM [Groups].[GroupMembers] AS [t6]
        ) AS [t7]
    WHERE [t7].[value] = 1
    )) = @p7)

So I started playing around with the LINQ statement until I reduced the size of my SQL query significantly. The following is the result of that optimization:

New Query

var groupMembership = (from gm in GroupMembers
                       let visibility = gm.Group.VisibilityPermission
                       where visibility == 'O' || (gm.IsApproved && gm.UserId == userId && (visibility == 'G' || (gm.RoleId & (int)Role.Manager) == 0))
                       select gm.GroupId).Distinct();

from i in Ideas
join xgl in GroupIdeaLinks on i.IdeaId equals xgl.IdeaId into groupLinksGroup
from gl in groupLinksGroup.DefaultIfEmpty()
where gl == null || groupMembership.Contains(gl.GroupId)
select i;

Which outputs the following SQL query:

SELECT [t0].[IdeaId], [t0].[CategoryId], [t0].[UserId], [t0].[IsPopular], [t0].[PopularOn], [t0].[Title], [t0].[SafeDescription], [t0].[Description], [t0].[Path], [t0].[Score], [t0].[Rank], [t0].[ExternalLink], [t0].[VideoLink], [t0].[BumpUpCount], [t0].[BumpDownCount], [t0].[TotalBumpCount], [t0].[TotalCommentCount], [t0].[CreatedOn], [t0].[rowversion]
FROM [Ideas].[Ideas] AS [t0]
LEFT OUTER JOIN (
    SELECT 1 AS [test], [t1].[GroupId], [t1].[IdeaId]
    FROM [Groups].[GroupIdeaLink] AS [t1]
    ) AS [t2] ON [t0].[IdeaId] = [t2].[IdeaId]
WHERE ([t2].[test] IS NULL) OR (EXISTS(
    SELECT NULL AS [EMPTY]
    FROM (
        SELECT DISTINCT [t3].[GroupId]
        FROM [Groups].[GroupMembers] AS [t3]
        INNER JOIN [Groups].[Groups] AS [t4] ON [t4].[GroupId] = [t3].[GroupId]
        WHERE (UNICODE([t4].[VisibilityPermission]) = @p0) OR (([t3].[IsApproved] = 1) AND (([t3].[UserId]) = @p1) AND ((UNICODE([t4].[VisibilityPermission]) = @p2) OR (([t3].[RoleId] & @p3) = @p4)))
        ) AS [t5]
    WHERE [t5].[GroupId] = [t2].[GroupId]
    ))

As you can tell the second query is much more compact and it does the exact same thing as the first query. I was pretty proud of my self and riding high on my genius, until this happened:

Well this didn’t really happen, but you get the point. I quickly came down from my high when I tested the performance of the new query in SQL Server Profiler, and received these results from the two queries:

• OLD QUERY (Reads 130, Durration 5)
• NEW QUERY (Reads 218, Durration 28)

That is right my optimization increased the number of times SQL has to read the table by 68% and time it takes to execute by 460%. So I reversed all my changes and learned a lesson on how not to optimize a LINQ statement.

The moral of the story is you probably don’t need to optimize your SQL query through LINQ, just keep it simple and optimize your LINQ statement and leave the rest up to the professionals at Microsoft who created the LINQ to SQL expression query generator.