Brig Lamoreaux

Automating Quality Code with FXCop

2009-05-22T21:05:00.001-07:00

FXCop is a tool that can analyze .NET manage code for conformance to design guidelines and your own custom rules. I recently found a great document that explicitly outlines how to create you own custom rules from Jason Kresowaty. Some possible custom rules I am considering include:

Security Rules

SQL Injection
Session Hijacking
Cross Site Scripting
Sensitive Data in Query String
Sensitive Data in Form Fields
Sensitive Data in ViewState

Performance Checks

YSlow Suggestions
Company Specific Rules
Using Common Logging

And More

Preventing Groupthink

2009-05-22T21:04:00.001-07:00

I’ve been looking into the idea of Groupthink and found a nice list on preventing it.

From Wikipedia:

Leaders should assign each member the role of “critical evaluator”. This allows each member to freely air objections and doubts.
Higher-ups should not express an opinion when assigning a task to a group.
The organization should set up several independent groups, working on the same problem.
All effective alternatives should be examined.
Each member should discuss the group's ideas with trusted people outside of the group.
The group should invite outside experts into meetings. Group members should be allowed to discuss with and question the outside experts.
At least one group member should be assigned the role of Devil's advocate. This should be a different person for each meeting.

Json Serialization

2009-02-03T19:42:00.002-07:00

I've been having fun trying to implement an idea I have. Part of the idea involves caching objects to memcached in Json format. The idea being that the object saved to memcached wouldn't be bound to .Net specific methodology.

What's out There

At first I looked into creating a Custom Formatter that would serialize my object for me. That seemed like an overkill so I searched for a faster and prepackaged library. After poking around to see what Json serializers were on the market, I found Json.NET.

At first this looked like a possible solution. Then I ran the Test Scripts on the Project and found a PerformanceTests test. This test script had timings for three different types of serializers. So my options were:

Performance

I changed the Json.Net Performance tests to serialize and deserialize 1000 times. The JavaScriptSerializer was the fastest during the tests.

Serializer	Time
JsonNet	203 ms
JavaScriptSerializer	95 ms
DataContractJsonSerializer	842 ms

Strangeness

One of the bad things I saw happen is how the DataContractJsonSerializer tried to serialize the private methods of the class. So I thought I would try to force it to use the properties of the class by not defining a backing variable.

public string Name { get; set; }

What to my surprise when I saw the following in the Json output:

"<Name>k__BackingField":"Rick"

Weird.

First Impressions of Google App Engine

2009-02-01T13:48:00.001-07:00

So I thought I would try to see what Google App Engine is all about. My main goal is to learn python a little better and also try to get a better understanding of how to build applications that can

Handle a lot of load
Be distributed
Have Standard Data access

Installing

I'm a .Net developer on a Windows machine and really don't have a lot of Linux/Python experience. I know I'm not alone so I am going to track my experience using GAE to build my car maintenance application know as Motozio.

I'm following the Google Getting started guide which is the following

Download and install Python 2.5
Download and install GAE

Coding

Installation is a breeze. The two python files for the hello world program are pretty straight forward too.

Trouble hit when I tried to test my application. I really didn't know how to call the script but I figured it out.

My application folder structure looks like this

c:\dev\helloworld\
c:\dev\helloworld\helloworld.py
c:\dev\helloworld\app.yaml

Running

I had to open a command prompt and type:

C:\dev>dev_appserver.py helloworld

Notice that the path to google_appengine, where the dev_appserver.py script resides, is already in the system path. The script dev_appserver wants the path to the application. Which is the helloworld directory.

Once I got that figure out. Bam! My helloworld was up and running at http://localhost:8080

I'm always impressed with out of the box examples of hello world. I really liked the ruby on rails example. And I am just as impressed with App Engine so far. Of course it doesn't take much to impress me.

Typing is Fun

2008-12-08T09:59:00.005-07:00

I've been reading about other developers taking typing test to see where they are. I thought I would jump in the game and post my results. I fit within the average typing speed between 50-70 wpm. Nothing to brag about. Although I did get a perfect score. Yes!

AjaxWorld - Day 3

2008-10-22T17:23:00.001-07:00

This last day started off being able to sleep in, but it quickly went downhill when the promised breakfast arrived 45min late. Nevertheless, I did glean a few bits of good information, and only had to walk out of a few sessions. I don't want to sound like a jerk, but some of the content of the sessions was more like an infomercial than a learning environment. Enough of my rant, on to the tidbits.

Social Networks as a Tool

We use social networks everyday to accomplish our jobs. How can we utilize the power of social networks in an computerized environment. Look at your job through the eyes of a social network. In other words, what pieces of information do individual's own.

Social networks can help me as a software developer produce higher quality of code.

How: Create tools that utilize the group. For instance, FxCop can produce a list of broken rules when it runs. These broken rules can be placed into a central repository. Then the entire group can access the rules and fix the code. The trick is to make it easy to select a rule, get the code, and make the fix. Bam. Almost like ants doing a little bit of work to get a big job done.
Groups are nodes. Treat group on the graph just like a regular node. It will have its own resources and connections and events. Groups can be used to manage projects
See Ning.

JQuery - Modular Code

A few people got their feelings hurt when Yehuda Katz changed his presentation from an introduction to JQuery to a more advanced talk on how to write modular code with JQuery. I enjoyed the talk albeit Yehuda rushed through the talk due to his 15 minute tardiness.

Use the Addon .listener, .intercept to bind events.
Bind to a higher element than lots of lower elements
Use Event Messaging
Think in High Level Events. Instead of e.keycode == 38 do .trigger('uparrow_clicked')
See LiveQuery. This allows you to bind events to all DIV even if they are not on the page yet. Useful for dynamic text via a widget
See

Become a good Presenter

Walking away from this conference it was easy to spot the great presenters from the lousy ones. Great presenters do the following:

Prepare quality Content
Format a clean unique presentation
Know the audience
Don't repeat the exact same general information someone else already covered.
Have quality content
Don't be an infomercial

The lousy presenters did basically the opposite

Show no emotion
Mumble
Appear as though you just saw the content
Put lots of text on each slide
Have misleading presentation titles in attempts to lure unsuspecting attendees
Have poor content

That's about it. I'm about to get on the plane and leave beautiful San Jose and look forward to putting to practice the real gems of the conference.

AjaxWorld - Day 2

2008-10-22T12:23:00.002-07:00

Another day and a little more information. This day had a couple of outstanding presentations. The first was by Kevin Lynch the CTO of Adobe. What a great presenter. The other was Douglas Crockford's survey of JavaScript.

General Tips and Tricks

JsLink to validate JavaScript

Yui Compressor for css/JavaScript minification

OpenAjax gives ajax guidelines

Use the # to maintain state within your app.

Push Data to the Cloud

Because users are starting to access the web through multiple devices, there is a need to modify your application to run in the different contexts. Kevin Lynch showed a music library application that accessed his person music through the computer, phone, and Wii.

Manage Large Data

Charles Kendrick had a lot of good ideas on how to sort and filter high volume dataset. The idea is similar to chunking data. When a sort needs to happen, push the sort back to the DB if the record count is larger than the client page limit.

Complete Server side Solutions

This section comprised three different sessions. These included Java Server Face, Google's GWT, and finally jMaki. The general idea is developers will use tools to create pages that can call back to the server. Karthik Shyamsunder showed how GWT will actually compile down the Java into JavaScript and also merge all images into a single image. The script created is specific for the browser and version.

One technique to do partial rendering of pages is to keep track of the DOM tree and perform a Diff between the two DOMs. Then send down the Diff file to have the client update the parts that need to be updated. Very clever.

Use a publish subscribe bus on the client for inter-component communication.

JavaScript the Good Parts

This was by far the most informational session. Douglas Crockford presented the ideas found in his book JavaScript: The Good Parts. In it, he showed how different he === and == operators are and how important it is to use the === all the time. He recommended using JsLint to verify your JavaScript to ensure you are following the best practices he found.

Variable User Experience

Mark Meeker from Obitz showed how they created their application to down grade gracefully. They first develop the page to work with POSH. Then depending on the user's browser, you can enhance the user experience by hiding detail with JavaScript and CSS. They also use build test to enforce their development standards. They make use of a composite view design. This is where you create many small controls that show specific data and have specific behavior. These controls are then assembled together onto a single page.

Next Steps

I think I would like to try writing an application in both GWT and Silverlight. I like the idea of being able to publish a single application have have it live on the network edge through a CDN. For now though, my apps will be widgets and JavaScript based.

AjaxWorld - Day 1

2008-10-21T07:15:00.001-07:00

First day over and I think the overall assessment is not too bad. The sessions started pretty early at 7:30 and went all the way till 8:00pm. That's ok if all the sessions had been stellar, but I would say they were only ok.

The day started off with Scott Guthrie showing off some Silverlight and MVC. Some of the talk highlights were:

The more api's the better. As more companies and application standardize and expose api's, the easier it is to consume those data and create mashups.
Business's need to think about multiple channels of delivering content. Users access the web via browsers, TVs, gaming consoles, their phones, and soon much more.

The rest of the sessions seemed to all blend into one. I'll try to take better notes on day two, but this is what I captured for the rest of the day,. Right away Dave Meeker said don't reinvent the wheel for social networks. Use third party frameworks and tools. OK. Now on to the big ideas that stood out.

Use 3D design

This means the z dimension on the page and also changing the size of elements according to their relevance. The examples I can think of is the office tool bar and the ITunes Music slider control. The control where on the edge of the screen the images seemed to be filed and stacked close together, but the center image is large and full. The idea is to put the most important content front and center, and push the rest to the sides.

Ajax Push / Comet.

Where have I been? I was blown away by this idea of the server being able to update content on the client. Ajax Push makes a lot of sense if you have many clients that are sharing a common resource that can change frequently. Ideas are, collaborative documents, messaging, and alerts.

Open Ajax

I was really excited about Open Ajax Alliance. This is a group that is attempting to create some standards around Ajax. The ideas that stood out to me were the OpenAjax Metadata Specification. This document goes about defining the term Widgets and Ajax api's. This is a great idea because now toolkits can be created that can read the information. Enterprise should focus on SOA widget end services. Most excitedly is they have a preview tool for widgets, although I forgot where. Widgets become valuable when they can talk to one another. I'm just not entirely sure how they do this.

Use Fiddler with .Net Application

2008-06-24T11:35:00.001-07:00

I often need to watch communication between a web service and my application. The best way I've found is to configure my app to use Fiddler as a proxy between my application and the web service. Once configured, I can watch the traffic. The simple step that I keep forgetting is the entry in the Web.Config file.

Get Fiddler Settings

Open Fiddler and click on Tools -> Fiddler Options. The Fiddler Options window will pop up and take note of Listen Port number under Proxy Settings. In this case the value is 8888.

Update Web.Config

Open your Web.Config file and add the entry. Make sure the port on the address matches the one defined in Fiddler.

<system.net>  <defaultProxy>    <proxy  proxyaddress="http://127.0.0.1:8888" />        </defaultProxy></system.net>

Other web refrenced talk about setting other attributes on the proxy element, but don't These seem to make it fail. Just the simple solution from Rick Strahl will make it work.

Logging Best Practices

2008-04-21T14:06:00.001-07:00

I just read a presentation from Network Intelligence on putting together a SIEM solution. I really don't know that much about SIEM, but what I do know is when my application breaks, I want to know what happened and have access to the logs as quickly and easily as possible.

I've talked about how to implement Logging in an application before, but this presentation sparked a few thoughts.

Don't Filter Logs at the Source

Log all events to a single place, then use a different tool to pull down the data to sort and filter.

Have a Reporting Time Period

Create a report that will run once a Day, once a week or whenever that shows all the events you are interested in seeing. After the report has ran, you can get rid of the data -or- you can archive it.

Standardized Time

Use a centralized, standard Time for every log. Synchronizing time between application logs will help correlate issues between applications.

Protect Sensitive Data

Be cautious of Sensitive information captured in the logs. Often username, machine names, and other sensitive data is captured in the event log. Provide access to the log to only those who need it.

Analyze Data

Every application should know what events will be logged during day to day usage. Create a baseline report and look for exceptions to the expected baseline.

Practical SVN Branching and Merging

2008-03-28T08:49:00.001-07:00

I love SVN and the ability to branch and tag my repository. I work on a two week release cycle with each revision being a version change. In SVN, I create a branch when I go into QA. Just about once a month I need to make a fix in QA which requires the need to merge a the branch fix into the trunk. Here are the steps I use in order to make that happen.

Steps to update Trunk with Branch changes

Finish and commit changes on branch.
Point working copy to trunk via Switch
Ensure working copy has latest changes via update
Merge changes of Branch in to Working Copy via Merge Command.
1. Click Merge
2. Start URL will be the Branch
3. Click Show Log to select the start of the branch
4. End URL will also be the Branch
5. Click the HEAD Revision. This will get all the changes made during the branch.
6. Take Note of the Start and End Revisions. These values will be used later. You will have to log at the log to see what the head revision is.
7. Click Dry Run to see what files will be affected.
8. If things looks good, click Merge.
Your working copy now has the changes made during the branch. Edit conflicts if any.
Commit you changes with the comment like: "Merged v2.0-branch changes r99026:99343 into the trunk"

I hope this helps, if not, I know it will help me because I keep forgetting how to do this. Let me know if you have a different approach of merging in patches.

Fiddler and Web Requests

2008-03-27T11:37:00.005-07:00

I often need the ability to watch the web traffic that goes on between webservice calls and web request my application makes. One way I've found is to use Fiddler to watch the traffic. The one important step I keep forgetting is how to tell my app to use the proxy. Here is the setting I need to make in the Config file. With this setting and Fiddler, I can now watch all the web traffic my application makes.

<system.net>
  <defaultProxy enabled="true">
     <proxy proxyaddress="http://127.0.0.1:8888/">
    </proxy>
  </defaultProxy>
</system.net>

ASP.NET Authorization - Out of the Loop

2007-11-14T15:29:00.001-07:00

The Problem

I have often created projects that have different areas restricted based on user roles. For instance, there might be a section only for administrator and another for power users then another for the general public. I needed a way to ensure users were restricted from accessing pages based on their roles. My solution worked and I thought it was pretty good, but I just realized there is a another way to do it.

My Way

I wanted to ensure that each page would perform a check to see if the current user was authorized to use the page. I created a base page which had an ApplyAuthorizationRules method which was kicked off during OnInit. By default I didn't do anything in the method so everyone was authorized to access the page by default.

public class BasePage : Page{    protected override void OnInit(EventArgs e)    {        base.OnInit(e);        ApplyAuthorizationRules();    }     protected virtual void ApplyAuthorizationRules()    {        // By default do nothing. Essentially allow access to everyone.    }}

Each page that needed authorization would derive from the base class and override the ApplyAuthorizationRules () method. Notice in the Class Diagram how the AdminPage is overriding the method while the other two pages, SummaryPage and DetailPage, use the default implementation.

The code in the AdminPage looks something like this:

public class AdminPage : BasePage{    protected override void ApplyAuthorizationRules()    {        if (this.User.IsInRole(Roles.Admin))        {            // Allow        }        else        {            Response.Redirect("~/Unauthorized.aspx");        }    }}

This method worked great and could handle all of our use cases when we needed to apply authorization. It worked so well I didn't dig any deeper into other solutions. To be honest I knew that ASP offered some tools for managing users and roles but I thought it was mostly tied to authentication. What to my surprise today when I put two and two together and realized there is an ASP way to handle authorization.

The ASP.NET Way

Ever since my first ASP.NET application I have generally used the web.config file to configure authentication. I have generally used something like this:

<system.web>  <compilation debug="true" />  <authentication mode="Forms">    <forms name="myAppAtuh" loginUrl="/login.aspx" />  </authentication>  <authorization>    <deny users="?"/>  </authorization></system.web>

I never really took a close look at the <authorization> tag. I can accomplish everything I was doing programmatically through configuration. With the following example, I'm able to create an Admin section on my app where only those users with an Admin role have access. There is a special case that allows users with a Role of 'Power' to see the /Admin/View.aspx page. You do have to be careful in creating the rules because they do execute in a top down manner.

<system.web>  <compilation debug="true" />  <authentication mode="Forms">    <forms name="myAppAtuh" loginUrl="/login.aspx" />  </authentication>  <authorization>    <deny users="?"/>  </authorization></system.web> <location allowOverride="false" path="/Admin/View.aspx">  <system.web>    <authorization lockItem="">      <allow roles="Power"/>    </authorization>  </system.web></location><location allowOverride="false" path="/Admin/">  <system.web>    <authorization lockItem="">      <allow roles="Admin"/>      <deny users="*"/>    </authorization>  </system.web></location>

Conclusion

My way isn't too bad. I do feel bad for not noticing that other people have talked about this before. Doing a simple search returns a lot of results showing everyone else already knew about the configuration method. Oh well, now I know, and you do too.

Beware Dynamic User Controls in a Web Farm

2007-11-02T16:01:00.001-07:00

Many large sites utilize a web farm with a load balancer. The load balancer will cycle through each server in the farm for each request. Today I assumed our load balancer had persistence. I think the text from Wikipedia sums my experience today: "But reliance on persistence can cause problems if the persistence fails." Come to find out, the rules on the load balancer were not setup uniformly and I found a special case where persistence failed.

My application was creating custom user controls dynamically on a form. The Page class offers two two ways to load a control dynamically:

public Control LoadControl(string virtualPath);public Control LoadControl(Type t, object[] parameters);

My code was using the second method. It stuffed the control's GetType().AssemblyQualifiedName property into the view state. Then on the PostBack, this information was retrieved from the ViewState to create a Type object and Load the control again.

One thing I didn't realize was the AssemblyQualifiedName used the Shadow copy assembly name, which is something like App_Web_XXXXXXXX where X can be any letter or number.

This would have been fine if persistence didn't fail on the load balancer. What happened however was the view state saved the type assembly name as App_Web_ob44m7cy.dll, which it was on Server1, but because the load balancer directed the request to Server2, no such file existed. And that was the error I was getting.

Type : System.IO.FileNotFoundException, mscorlib, Version=2.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089 Message : Could not load file or assembly 'App_Web_ob44m7cy, Version=0.0.0.0, Culture=neutral, PublicKeyToken=null' or one of its dependencies. The system cannot find the file specified. Source : mscorlib

The fix was two fold, tell the network team they didn't have uniform rules in place, shame shame shame. Second, don't rely on persistence being setup on the load balancer. I changed my code to load the control via the path. A pretty simple solution for a strange situation.

Getting Started with Lisp

2007-10-23T10:21:00.001-07:00

Lisp is an old and powerful language that influences modern languages like C#. Lisp is also difficult to jump into. I'll outline where to go to download a compiler and the steps needed to start using Lisp today.

Lisp is known as the programmable programming language. I became interested in Lisp when I learned how many of the sexy features in C#, like anonymous delegates, were implemented in Lisp 20 years ago. I wanted to find out more about Lisp but I didn't know where to start. The first step I found is to setup a Lisp environment.

Download a Compiler

Lisp has many different dialects. Common Lisp is one dialect. There are several different implementations of Common Lisp similar to the different implementations of C++. The first step is to choose and download a specific Lisp compiler. I decided on Steel Bank Common Lisp (SBCL) for no rhyme or reason. One hurdle to overcome is understanding the unorthodox grid of available downloads.

The secret to the grid is to look for your OS in left column and your processor in the top column. I was directed to the yellow square labeled 1.0.9. This links to a installation file called sbcl-1.0.9-x86-windows-binary.msi. Executing the installer will install some files to the in C:\Program Files\Steel Bank Common Lisp\1.0.9\ folder. The most important file is the sbcl.exe. I actually created a shortcut to this file and placed it in my C:\Documents and Settings\All Users\Start Menu\Programs\Lisp\ folder

Stand your Ground

Running the sbcl.exe executable and firing up Lisp for the first time is a little daunting. You'll get a sappy message about a Kitten of Death and then just a * prompt, almost mocking you saying 'Here's your Lisp environment...now what are you going to do?"

Just shoot back with this and press enter:

(+ 2 3)

Lisp will reply with:

5

Lisp is conquered. I now feel confident in taking the next step and actually studying the language to see what it can do. For this I highly recommend Practical Common Lisp.

Future Plans

I like the idea of being able to pass around functions and redefine them on the fly. I feel similar to Lawrence and appreciate how clean the language is. It is void of a lot of keywords. I don't have any plans of writing an entire application in Lisp, but if I run into needing a small utility app, I will definitely consider using Lisp.

How to Convert Integers to Strings in C#

2007-08-31T12:21:00.001-07:00

I've been working with converting strings to integers and have a few options at my disposal. In the past I've bounce between using Convert.Int32(string) and Int.Parse(string). I've decided to look into what really goes on under the covers between these different options.

There are three different ways to convert one item to another:

Implicit Conversion Operator
Explicit Conversion Operator
Use a Format Provider

Conversion Operator

In order for the implicit and explicit conversions to work, the class/structure must define a Conversion Operator. The example below will allow an explicit conversion from int to SampleClass. Notice how the method takes an int and returns the SampleClass type:

class SampleClass{    public static explicit operator SampleClass(int i)    {        SampleClass temp = new SampleClass();        // code to convert from int to SampleClass...         return temp;    }}

Using the conversion in real code would then be:

int i = 45;SampleClass sc = new SampleClass();sc = (SampleClass)i; // Explicit conversion

If I had defined the method as implicit instead, then I wouldn't need the cast like operation. I could have just written:

sc = i; // Implicit conversion

What I really wanted to know is how strings are converted into integers. The String class doesn't implement either an implicit or explicit conversion to int. This leaves me the option of having to use a Format Provider object.

public class String{    static public explicit operator int(String string) // Doesn't exist.    static public implicit operator int(String string) // Doesn't exist.}

Format Provider

The framework offers two Format Provider methods:

Convert.ToInt32(string)
Int.parse(string)

The two are actually connected because Convert.ToInt32 calls Int.parse after it does a check for a null string.

Int.parse() is the real worker because it dives deep into several methods and finally does all the work to convert text into different numbers.

Int.parse(string)
- calls Number.ParseInt32(s, NumberStyles.Integer, NumberFormatInfo.CurrentInfo)
  - calls Number.StringToNumber(s, style, ref buffer1, info, false)
    - Calls Number.ParseNumber(ref chPtr2, options, ref number, info, parseDecimal)
      - Does all the work here

Summary

There are no implicit or explicit conversions from a string to an integer, you must use a format provider method. The two options available with the library are really one option.

	Is Null	Is Not Number
Convert.ToInt32(string)	Return 0	FormatException
Int.Parse(string)	ArgumentNullException	FormatException
Int.TryParse(string, out int)	Return false	Return false

Review of Copy Source As Html Tools

2007-07-19T12:31:00.001-07:00

It makes sense that as a blog about .net programming I should include samples of code. Code is easier to read when there is syntax highlighting and proper indentation. I'm going to look at four tools that provide a way of displaying source code in Html documents. The first tool is a general add in to Visual Studio and the last tools are specific to Windows Live Writer.

The contestants are the following:

CopySourceAsHtml

Insert Code for Windows Live Writer

Paste from Visual Studio

Code Snippet plugin for Windows Live Writer

The plugins for Live Writer really are not named very well, but I'll give them a shot anyway.

Screen Shot

In order to do a fair test, I will be comparing how the different tools handle the code in this screen:

CopySourceAsHtml

This is an add-in for Visual Studio. Right clicking on the code to copy displays a Copy As Html... option in the context menu. Once clicked, an option box appears which provides additional configuration. I've used this tool in the past and have always been satisfied with the results. I selected Wrap words and Embed styles as the only options for the copy which produce the following results:

using System;
 
/// <summary>
/// Wizard represents a humanoid with special abilities
/// </summary>
public class Wizard
{
    const string Gretting = "Greetings and well met.";
 
    /// <summary>
    /// Says the greeting.
    /// </summary>
    public void SayGreeting()
    {
        // Ensure it is Daylight Saving Time
        if (DateTime.Now.IsDaylightSavingTime())
        {
            Console.Write(Math.Abs(Math.PI + 25.4));
        }
        else
        {
            Console.Write(Gretting);
        }
    }
}

</Example>

Embedding the styles into the code gives the advantage of preserving the color highlighting when your post is picked up by an aggregator. If you do not embed the styles then you will lose the formatting and the code appears as normal text.

The tools provides options that really allow you to produce exactly what you want.

The only bad thing is I am forced to switch to html mode in order to insert the code snippet.

Insert Code for Windows Live Writer

This is the first tool I've used as a plugin for Windows Live Writer. After the installation, I see a new option in my Insert list for Insert Code...

The web site for Insert Code describes it as able to format a snippet of text in a number of programming languages such as C#, HTML, MSH, JavaScript, Visual Basic and TSQL.

Here is the results of the insert:

using System;

/// <summary>
/// Wizard represents a humanoid with special abilities
/// </summary>
public class Wizard
{
    const string Gretting = "Greetings and well met.";

    /// <summary>
    /// Says the greeting.
    /// </summary>
    public void SayGreeting()
    {
        // Ensure it is Daylight Saving Time
        if (DateTime.Now.IsDaylightSavingTime())
        {
            Console.Write(Math.Abs(Math.PI + 25.4));
        }
        else
        {
            Console.Write(Gretting);
        }
    }
}

</Example>

Right away I can see that the color for strings is different that what I have defined in the original. More importantly I see I lost the specific font I had defined. The biggest advantage over CopySourceAsHtml is not forcing me to switch to html mode in order to insert the code. The plugin makes it completely transparent to me.

I think I'm going to run into trouble on some aggregators when it goes to process the code. Insert Code creates a <style> </style> section where it defines the css classes used to format the code. The tool doesn't provide the option for me to select embedded styles nor does it give me options to tweak the output.

One thing I do like is the ability to turn on alternate line background.

Paste from Visual Studio

This is also an plugin for Windows Live Writer and also has nearly the same number of downloads as Insert Code. From the description on the web page, it can easily transfer syntax highlighted source code from Visual Studio to elegant HTML in Windows Live Writer. Copy from Visual Studio and insert directly to Windows Live Writer to maintain your unique syntax highlighting settings.

using System;

/// <summary>
/// Wizard represents a humanoid with special abilities
/// </summary>
public class Wizard
{
    const string Gretting = "Greetings and well met.";

    /// <summary>
    /// Says the greeting.
    /// </summary>
    public void SayGreeting()
    {
        // Ensure it is Daylight Saving Time
        if (DateTime.Now.IsDaylightSavingTime())
        {
            Console.Write(Math.Abs(Math.PI + 25.4));
        }
        else
        {
            Console.Write(Gretting);
        }
    }
}

</Example>

Sure enough, this tool has kept the colors and font I have defined in VS. The tool doesn't give me any type of user interface but takes the contents of the clipboard and inserts them as html. I'm guessing the tool uses the rich text formatting that VS provides and builds the styles. I do like that the styles are embedded in the code so I won't lose any formatting in the aggregators. However, I miss having the opportunity to add any tweaks to the output before it is committed.

Insert Code Snippet

The last Live Writer plugin to review is Code Snippet plugin for Windows Live Writer. This follows the same installation as the previous plugins. However, I do get the chance to make some changes before the code is inserted into my page.

I stick with the defaults and get the following results:

   1: using System;   2:     3: /// <summary>   4: /// Wizard represents a humanoid with special abilities   5: /// </summary>   6: public class Wizard   7: {   8:     const string Gretting = "Greetings and well met.";   9:    10:     /// <summary>  11:     /// Says the greeting.  12:     /// </summary>  13:     public void SayGreeting()  14:     {  15:         // Ensure it is Daylight Saving Time  16:         if (DateTime.Now.IsDaylightSavingTime())  17:         {  18:             Console.Write(Math.Abs(Math.PI + 25.4));  19:         }  20:         else  21:         {  22:             Console.Write(Gretting);  23:         }  24:     }  25: }

</Example>

Not too bad. In fact, besides not keeping my original color and font, I like the end result. I especially like how the styles are embedded inline. I don't think I would use the line numbers too often but I do like the alternate line background.

The Results

Among all of the tools I will keep two. I will keep CopyCodeAsHtml so I can get my code into html format whenever I want. For Windows Live Writer, I'm going to stick with the aptly named Code Snippet plugin for Windows Live Writer. I chose this as the winner because I am provided options over the results and it has inline styles. Insert Code for Windows Live Writer gives me almost the same features but doesn't provide inline styles which is a must when it comes to aggregators.

I do wish that more languages were supported in the plugins and hope this will come in time. I know there must be more tools out there to get source code to html but these are the four that I found. If you have a tool that you enjoy, please share it. I'm always looking for ways to improve and make life easier.

How to Implement logging

2007-07-18T15:33:00.001-07:00

We are getting ready to send our first child off to kindergarten and I wish I could be a fly on the wall to see how he enjoys his first day of school. Secretly I want to make sure he remembers the one-two punch I taught him in case someone tries to bully him around. Fortunately at work I have much more control over keeping track of my applications once I send them into production. The secret is Logging.

Our team has used three major logging implementations. Log4Net, Enterprise Library, and a homegrown solution. What we have discovered is the importance of making the implementation transparent to our application.

The Logging Wrapper

The wrapper class minimizes the impact to your application when you switch out a logging implementation say from Log4Net to Enterprise Library Logging. I have found the following interface to completely accommodate our needs.

public interface ILogger
{
    void LogInformational(string message);
    void LogWarning(string message);
    void LogWarning(Exception ex);
    void LogError(Exception ex);
}

I use three categories for the messages specifically so I can turn on and off logging by setting a threshold level.

Informational: These messages will be used liberally throughout the code to provide process detailed information similar to debug traces.
Warning: Warnings are events the application can recover from. These include messages that provide more detail when the application responds to the user "Unable to save record". Warnings also include all exceptions that are handled in a try catch block.
Errors: Errors are exceptions that are logged from the global error handler.These are errors that the application did not expect.

The Implementation

We use Enterprise Library for .NET Framework 1.1, June 2005 in our projects. We are still evaluating Enterprise Library 3.0 – April 2007 (for .NET Framework 2.0 and 3.0). In the meanwhile this is how I configured EntLib.

First I include the following dlls

Microsoft.Practices.EnterpriseLibrary.Common
Microsoft.Practices.EnterpriseLibrary.Configuration
Microsoft.Practices.EnterpriseLibrary.ExceptionHandling
Microsoft.Practices.EnterpriseLibrary.ExceptionHandling.Logging
Microsoft.Practices.EnterpriseLibrary.Logging

Then I create a class that implements ILogger and use EntLib to get the job done.

using System;
using Microsoft.Practices.EnterpriseLibrary.Logging;
using Microsoft.Practices.EnterpriseLibrary.ExceptionHandling;
 
public class EntLib11Logger : ILogger
{
    public void LogInformation(string message)
    {
        Logger.Write(message, "Information", 1);
    }
    public void LogWarning(string message)
    {
        Logger.Write(message, "Warning", 5, 1, Severity.Warning);
    }
    public void LogWarning(Exception ex)
    {
        ExceptionPolicy.HandleException(ex, "Warning");
    }
    public void LogError(Exception ex)
    {
        ExceptionPolicy.HandleException(ex, "Error");
    }
}

Notice how both the Enterprise Library Logging and Exception Handling is used to provide functionality for our ILogger. With the Logging Block, I have to specify in the code the Priority level and Severity level. The Exception Handling Block allows those values to be set in the configuration file, which is what needs to be configure now.

Configuration

Configuration of Enterprise Library can be a little confusing but lets just go one section at a time.

The first thing we'll want to setup is the logging aspect. This makes the Exception Handling configuration a lot easier. Start up the EntLibConfig.exe tool and select File - Open Application and select the app.config file for the project.

Logging Configuration

Select your app node and choose to add the Logging and Instrumentation Application Block.

The tool allows you to define categories, formatters, and sinks. Sinks are the final resting place for our messages. Sinks can be a flat file or an event log. Formatter are used to turn our message object into a plain text message. The category is where you define what happens to our message object when it is caught.

I modify the Sinks section to have one sink for the Application Event Log.

Modify the categories section to have the three categories defined during the implementation step

Information
Warning
Error

Notice how each category has a single destination. The destination tells Enterprise library to use the default Text Formatter and send it to the Application Event Log Sink.

One last step in logging is to set up a default category. I like to set mine to default to Error.

At this point your application should be able to use LogInformational(string message) and LogWarning(string message)with no problems. Setting up the exception handling will take care of the rest.

Exception Handling Configuration

Select your app node and choose to add the Exception Handling Application Block. Select the application block and add a new Exception Policy with the name Error. Add a new Exception Type to the policy that catches all Exceptions. Finally add a new Logging Handler.

The configuration I use for the Logging Handler of the Error Exception Policy is:

LogCategory	Error
Priority	10
Severity	Error

The Warning exception policy is exactly like the Error but with the following values for the Logging Handler:

LogCategory	Warning
Priority	5
Severity	Warning

With both exception policies setup and using the logging our ILogging class is fully functional. Now it is a simple configuration setting to turn on and off the different logging levels.

Any value larger than the MinimumPriority will be logged. Anything less will not be logged, simple. The values of the categories are: Error-10, Warning-5, and Information-1. I like to set the MinimumPriority to 10 in production.

Wrap Up

So far this method of logging has been very useful. There is a lot more you can do from here such as sending emails when a warning happens. I've only laid the foundation of a logging methodology.

In the future I plan to evaluate how the Enterprise Library 3.0 – April 2007 version handles logging and create an implementation for ILogger. Other things I've thought about is to create some sort of LoggingFactory or ServiceProvider to really disconnect my application from the implementation. Ideally I don't want to have any referenced dlls in my apps.

Tell me what your experience has been with logging. What works well for you, what doesn't. Do you see yourself using the ILogger interface. Please let me know because I would love to improve on what I have.

When a Bug is Found

2007-07-17T08:36:00.001-07:00

James Shore talks about the concept of being bug free and has many recommendations on how to reach that goal. At my desk I have the following guide hanging on the wall to help me in my quest.

First:

Create a Test to Prove it:
Fix it:

You've fixed the problem, but you haven't solved the underlying cause.

Discover why that bug occurred.

Discuss the code process with your pairing partner.
Is there a design flaw that made this bug possible?
Can you change an API to make such bugs more obvious?
Is there some way to refactor the code that would make this kind of bug less likely?
Improve your design.

If you've identified a systemic problem, discuss it with the rest of you team in you next meeting or iteration retrospective.

Tell people what went wrong so they can avoid that mistake in the future.

Ruby Array Randomize

2007-07-11T12:36:00.001-07:00

I needed the ability to randomize the order of an array. This is useful when you have a set collection of questions in a survey that you want to shuffle before you present them to the user. After looking around a bit, I didn't find anything that worked. Here is what I pieced together with my limited knowledge of Ruby.

# Created by Brig R. Lamoreaux

# Date: 11 July 2007

# Time: 10:38 am

# Randomize the contents of an Array

class Array

def randomize

a=self.dup

result = []

self.length.times do

result << a.slice!(rand(a.length))

end

return result

end

def randomize!

a=self.dup

result = []

self.length.times do

result << a.slice!(rand(a.length))

end

self.replace result

end

Here are some test that make sure it really works.

puts "Randomize Tests"

x = [1, 2, 3, 4, 5]

print "x: "

x.each do |i|

print "#{i}"

end

y = x.randomize

print "\ny: "

y.each do |i|

print "#{i}"

end

x.randomize! # x is now [3, 5, 4, 1, 2]

print "\nx: "

x.each do |i|

print "#{i}"

end

x.randomize! # x is now [3, 5, 4, 1, 2]

print "\nx: "

x.each do |i|

print "#{i}"

end

Special thanks to syntaxhighlighter on providing styles for ruby syntax highlighting.

Update: After doing a search for array shuffle, I found some useful examples.

Balance Simple Code and Expert Knowledge

2007-07-05T10:36:00.001-07:00

I enjoyed reading What’s Wrong With This Code (#16) because it got me thinking about having in depth knowledge on a subject and KISS. When I saw the code, I thought it was an overkill for the problem. In the comments I offered a solution that is simple and what seems to have solved the problem. I'll admit that I'm not an expert on JavaScript and I wouldn't be surprised if someone pointed out serious flaws with my code. While I make my way on the path to becoming an expert, I'll continue to solve problems with simple solutions. Borrowing the concept Red/Green/Refactor, from TDD, I think it is important for developers to follow the KISS/Learn/Revise model of development. It's impossible for me to become an expert in every field. But I should be trying to become an expert in the fields that I deal with. Just like eating an elephant, I needs to learn the technologies one piece at a time. The easiest way I do that is to glean from other's insight. It is amazing how much useful information is shared through blogs. I make it a practice to read the thoughts of the leaders in the industry several times a week. More importantly I am sharing my own thoughts. I recently searched my name and found that other people have found some of my solutions useful.

In answer to how to become an expert while keeping things simple, my solution is slow and steady, one article at a time.

Keeping the Blog Alive

2007-07-02T09:34:00.001-07:00

My daily routine involves reading several different blogs as my way of taking advantage of other people's great ideas. Most of the things that I read are technical, but mix in with the bunch has been a blog about blogs. For the most part many of the articles I read are in essence a variation on the same theme. It is refreshing to find a diamond in the rough article that really helps to clarify an idea or answer a difficult question. I don't expect this post to be a diamond, but I do plan for it to help me stay consistent with delivering my own ideas and experiences. The one piece of advice I've read is to keep a regular schedule of posting and that is what I'll do.

Objects, Data, and Actions

2007-06-22T21:30:00.001-07:00

I was recently asked what objects are in programming and how a database is related to them. Thinking long and hard about it, I came up with the following example:

Objects are containers that stores data and code. In fact objects are just like real life objects. Imagine an Animal object. Pick any animal you like, it will have Data and Actions. The data would be things like Height, Length, Age, Name, Color. And Actions could be Speak, Walk, Jump, Eat.

So a Lion could be:

MyLion.Height = 4

MyLion.Length = 7

MyLion.Age = 4

MyLion.Name = Mufasa

MyLion.Color = Yellow and Brown

Notice how these values can easily be stored in a database. But what about the actions? When MyLion performs an action, something gets done.

MyLion.Speak -> from my speakers I hear a "ROAR!!!!"

MyLion.Eat -> Now MyLion.Length = 7.5

See how actions can't really be stored as a value because they do something. A relational database can store all of my Properties but can't store what will happen when my object performs an action. But an OODB database can store objects. So I could save MyLion right to the database. I could also save MyChicken and MyHippo. Each object would do something a little different when I ask it to Speak or Eat.

Hope this helps.

Live Writer

2007-06-14T08:45:00.001-07:00

I've downloaded Windows Live Writer to see how well it will do posting to Blogspot.com, and so far I am very impressed with the interface. One of the features of Live Writer is the ability to select View -> Web Preview. This shows me what the current post will look like using the templates and styles I have set up at Blogspot.

Anonymous Delegates

2006-12-29T16:05:00.000-07:00

I never thought I would use anonymous delegates mostly because I didn't fully understand what benefit they could give me. If I can't get something out of a new tool or methodology, why use it. With anonymous delegates I can get a lot of great things. Specifically simple solutions to my problems

Problem

What is the best way to add filtering capability on a specilized collection? After looking at related problem sets, I found the public T Find(System.Predicate<T> match) method which is implemented on System.Collections.Generic.List. A Predicateis a special delegate that takes in an object, and returns true. Suppose I have a collection of names and I would like to check if a name is long. I could do the following

 class Program 
    { 
         static void Main(string[] args) 
         { 
             List<string> names = new List<string>(); 
   
             names.Add("Mary"); 
             names.Add("George"); 
             names.Add("Hank"); 
             names.Add("Billy"); 
             names.Add("Susan"); 
             names.Add("Ed"); 
   
             Console.WriteLine(IsLongName(names[4]));
 
         } 
   
         static bool IsLongName(string name) 
         { 
             return name.Length > 4; 
         } 
    } 

I define a method that checks if the length of the name is greater than 4. Well my method IsLongName is in fact a System.Predicate. Since the Find() method takes a Predicate as the match pattern. I can pass this method to the Find() method and get a collection of just long names. Let's use Find() to get a list of names that are long names. First I'll need to wrap the IsLongNamefunction inside of a delegate. Then I can pass that delegate into the Find() method.

 class Program 
    { 
         static void Main(string[] args) 
         { 
             List<string> names = new List<string>(); 
   
             names.Add("Mary"); 
             names.Add("George"); 
             names.Add("Hank"); 
             names.Add("Billy"); 
             names.Add("Susan"); 
             names.Add("Ed"); 
   
             // WrapIsLongName up insided of Predicate delegate. 
             Predicate<string> longNameMethod = new Predicate<string>(IsLongName); 
   
             List<string> longNames = names.FindAll(longNameMethod); 
   
             Console.WriteLine("There are {0} long names", longNames.Count); 
             Console.ReadLine(); 
         } 
   
         static bool IsLongName(string name) 
         { 
             return name.Length > 4; 
         } 
    } 

The output is: There are 3 long names. What if I need to get a list of just names that start with the letters M Gand B? No problem, create a new method that matches the signature of a Predicate and you are in business.

 class Program 
    { 
         static void Main(string[] args) 
         { 
             List<string> names = new List<string>(); 
   
             names.Add("Mary"); 
             names.Add("George"); 
             names.Add("Hank"); 
             names.Add("Billy"); 
             names.Add("Susan"); 
             names.Add("Ed"); 
   
             // WrapIsLongName up insided of Predicate delegate. 
             Predicate<string> longNameMethod = new Predicate<string>(IsLongName); 
             Predicate<string> mgbName = new Predicate<string>(IsMGBName); 
   
             List<string> longNames = names.FindAll(longNameMethod); 
             List<string> mgbNames = names.FindAll(mgbName); 
   
             Console.WriteLine("There are {0} long names.", longNames.Count); 
             Console.WriteLine("There are {0} MGB names.", mgbNames.Count); 
   
             Console.ReadLine(); 
         } 
   
         static bool IsLongName(string name) 
         { 
             returnname.Length > 4; 
         } 
   
         static bool IsMGBName(string name) 
         { 
             if(name.StartsWith("M") || name.StartsWith("G") || name.StartsWith("B")) 
             { 
                return true; 
             } 
             returnfalse; 
         } 
    } 

The console shows: There are 3 long names.There are 3 MGB names.

Clean it up

Now I see how powerful and convenient it is to be able to pass in a method that contains the logic to find an item in a list. But let's see how anonymous delegates can help out. All that an anonymous delegate gives us is the ability to create our methods on the fly; no need to have random methods sitting around that won't be used.

 class Program 
    { 
         static void Main(string[] args) 
         { 
             List<string> names = new List<string>(); 
   
             names.Add("Mary"); 
             names.Add("George"); 
             names.Add("Hank"); 
             names.Add("Billy"); 
             names.Add("Susan"); 
             names.Add("Ed"); 
   
             // WrapIsLongName up insided of Predicate delegate. 
             Predicate<string> longNameMethod = newPredicate<string>(IsLongName); 
             Predicate<string> mgbName = newPredicate<string>(IsMGBName); 
   
             List<string> longNames = names.FindAll(delegate(string name) { return name.Length > 4; }); 
             List<string> mgbNames = names.FindAll( 
                delegate(string name) 
                { 
                    if (name.StartsWith("M")|| name.StartsWith("G") || name.StartsWith("B")) 
                    { 
                        return true; 
                    } 
                    return false; 
                }); 
   
             Console.WriteLine("There are {0} long names.", longNames.Count); 
             Console.WriteLine("There are {0} MGB names.", mgbNames.Count); 
   
             Console.ReadLine(); 
         } 

The console shows: There are 3 long names.There are 3 MGB names.

Adding Parameters

What if my special search needs an additional parameter? Just create a wrapper class

 class Program 
    { 
         static void Main(string[] args) 
         { 
             List<string> names = new  List<string>(); 
    
              names.Add("Mary"); 
              names.Add("George"); 
              names.Add("Hank"); 
              names.Add("Billy"); 
              names.Add("Susan"); 
              names.Add("Ed"); 
    
              Predicate<string> isNameBilly = IsName("Billy"); 
              List<string> billyNames = names.FindAll(isNameBilly); 
              Console.WriteLine("There are {0} Billy names.", billyNames.Count); 
    
              Console.ReadLine(); 
          } 
    
          static  Predicate<string> IsName(string name) 
          { 
              return  delegate(string item) 
              { 
                 return item.Equals(name); 
              }; 
          } 
     } 

I can create a method that returns a method. What is amazing about this is the ability to pass extra parameter into the dynamic method.