I'll be giving a talk on implementing Speech Recognition in .NET this Saturday.  There will be 24 presentations delivered at the code camp.  Register to attend the code camp here.

Adding Speech Recognition to your Application

Since version 3.0, the .NET framework has built in support for speech recognition. It is easy and straightforward to use. You will learn how to use the speech recognition capabilities of .NET. We'll talk about the capabilities of the speech recognition classes. And we talk about best practices and lessons learned.

There really is a single instance of a static field per appdomain.  It kind of says this in the C# language reference, but not very explicitly.  This knocked me around a little because I had assumed that there was a single static field per type within an appdomain.  I had a base type that had a dictionary as a static field and assumed that each class that derived from it had its own instance of the static field.  Not so…

Look at the following test code:

using System;
using System.Collections.Generic;
using System.Text; 

namespace ConsoleApplication3
{
    class Program
    {
        static void Main(string[] args)
        {
            baseClass.AStaticValue = 3;
            inheritedClass.AStaticValue = 4;
            Console.WriteLine("baseClass {0}", baseClass.AStaticValue);
            Console.WriteLine("inheritedClass {0}", inheritedClass.AStaticValue);
            Console.ReadLine();
        }
    }
 
    public class baseClass
    {
        public static int AStaticValue = 0;
    }

    public class inheritedClass : baseClass
    {

    }
  }
 }

I expected the results to be
baseClass 3
inheritedClass 4

Instead, the output is
baseClass 4
inheritedClass 4

So unless you decorate the field with ThreadStatic or ContextStatic attributes, static means just one in an appdomain.  Remember that when you are inheriting from objects that have static fields, there’s only one instance of that static field.

Part of the .NET 3.0 release was the System.Speech namespace.  This gives you easy access to powerful speech capabilities.  Text to Speech, Speech Recognition, and Dictation are all part of this library.  I've tried speech recognition every 5 years or so, and it has always worked, but it never allowed me to work faster than just using my keyboard.

As a consultant, I work on a variety of different machines.  Sometimes just my laptop, sometimes my laptop with an external monitor, and other times a dual monitor system.  Pictured below is how I set it up on my laptop. 

Most panels are auto hide so that I can maximize the amount of space to view the code.  Getting to the solution explorer means I have to take my hands off the keyboard, and go to the mouse to select the tab.  There are keyboard shortcuts, but I either don't remember all of them, or they don't work in all types of documents.

Since speech recognition is so easy to use with the system.speech dll, and Visual Studio addins aren't rocket science either, I built an addin that enables speech recognition within Visual Studio.  Basically, the addin maps a list of words, that when recognized, execute a command in the Command Window.  So, when I need to quickly look at the task list, I don't have to remember the keyboard shortcut, or mouse to it.  I just speak "Task List".  When I need the toolbox, I say, "Toolbox".  I didn't get into dictation, because I don't think it is efficient to speak "for space open parenthesis int i space equal space zero semicolon i less than items period count semicolon i plus plus close parenthesis open curly brace..."

I uploaded the source and an installer here on CodePlex.  Because (I think) Visual Studio 2005 only works with addins written in .NET 2.0, this only works for VS 2008.  Try it out and let me know how it works.

Me either.  See here in the C# Language Reference.  This means you can overload the operators for your classes.  The example on MSDN talks about classes that have a true, false, or null (neither true nor false) state.  You could also create a type that can be both true and false.

Overloading operators can lead to code that is difficult to read.  You should only overload operators when it makes sense to do it.  If it is unclear what “if (myType)” means, don’t overload the operators.

I wrote a small rules engine for a company a while back and this would have been helpful.  Supplemental rules were implemented as separate classes so that they could be dynamically strung together in different orders as the business logic changed.

Here’s a contrived sample using the true/false operators:

    public class Customer
    {
        public string CustomerName { get; set; }
        public string EmailAddress { get; set; }
        public decimal OutstandingBalance { get; set; }
    }

    public class CustomerIsGoodRule
    {
        
        public CustomerIsGoodRule(Customer customer)
        {
            this.Customer = customer;
        }

        Customer Customer { get; set; }

        public static bool operator true(CustomerIsGoodRule theCustomer)
        {
            return (theCustomer.Customer.OutstandingBalance <= 0);
        }

        public static bool operator false(CustomerIsGoodRule theCustomer)
        {
            return (theCustomer.Customer.OutstandingBalance > 0);
        }
    }

    class Program
    {
        static void Main(string[] args)
        {
            Customer c = new Customer();
            c.CustomerName = "Joanne Doe";
            c.EmailAddress = "joannedoe@mailinator.com";
            c.OutstandingBalance = 25;

            CustomerIsGoodRule cigr = new CustomerIsGoodRule(c);

            if (cigr)
            {
                SendNewCatalog();
            }
            else
            {
                SendNewStatement();
            }
        }
    }

One thing to keep in mind is that if you provide a definition for true, you must also provide one for false.  Also notice that the logical negation operator (!) was not overriden, so a statement like if (!cigr) fails to compile.

I haven’t taken the time to upload the materials yet, so thanks to Steve Coombes for hounding me about it.  Here are the files:

Powerpoint Presentation

Source Files/Demo

In case you haven't heard...  Notice the new location in Nashua.

New Hampshire Dot Net Users Group is back just in time for Microsoft’s latest releases of Visual Studio

We’ll be meeting In our new home at

6:00-8:30 May 15th

Eaton Richmond Center room 100

Daniel Webster College

The Guest Speakers will be

Phil Denoncourt

Overview of the AJAX control toolkit:

AJAX is an way to make websites more interactive and responsive.  It's powerful and easy to use.  The AJAX control toolkit is a set of controls that use AJAX and wrap common UI scenarios.  Cascading dropdowns, Auto completing textboxes, in-page popups, hovering menus and dynamically populated textboxes are some of the controls we'll go over.  No AJAX experience is required.

And Pat Tormey will be presenting

What’s new and important in VS 2008 with a little bit of WPF and Expression Suite

Stay tuned to http://www.nhdn.com for details and register up for notices…

I'm going to be taking this next week, so I thought I'd post my study notes as I assemble them.

• May include but is not limited to: using the ConnectionStringBuilder; leveraging the ConfigurationManager; protecting the connection string; using Security Support Provider Interface (SSPI) or SQL Server authentication; correctly addressing the SQL Server instance; managing “User Instance” and AttachDBfilename

The ConnectionStringBuilder objects allow for an abstract way of building a connection string for ADO.NET.  There is a good writeup on the data access blog showing basic usage.  There is a SqlConnectionStringBuilder, OleDbConnectionStringBuilder, OdbcConnectionStringBuilder, and an OracleConnectionStringBuilder.

The ConfigurationManager has a specific section for storing connection strings and the ConnectionStrings property to manipulate them.  See this article on MSDN for more information.  An area that doesn't get much attention is that the Connection string configuration can be stored in a separate config file.  See the section titled "Using external configuration files" to see how to use the configSource property.

MSDN has an article on securing configuration strings.  An important part to pay attention to is encrypting connection string information stored in config files.  Scott Forsyth has a post on using aspnet_regiis to protect web.config files.  Daruish Tasdighi has an article at codeproject about protecting the contents using the ConfigurationSection.ProtectSection / UnprotectSection methods.

• Manage connection objects.
  May include but is not limited to: managing connection state, managing connection pool; implementing persistent data connections; implementing Multiple Active Result Sets (MARS); encrypting and decrypting data

Managing connection state.  I think that is knowing that there are methods on the connection object to open and close the connection.  The connection will be closed when the connection object is disposed.

There is a great post by Sijin Joseph about connection pooling with all the different connection objects.  I wasn't aware that there is no connection pooling with ODBC using the managed providers.  Also interesting was that connection pooling for SQL Server was disabled while debugging in Visual Studio.  Another thing to be aware of is that when using Integrated Security for SQL Server, the domain/user is added to the list of things that need to match for a connection to be reused.

To have a persistent data connection, you disable connection pooling and keep the instance of the connection object around.  The connection will remain open until the connection object is disposed or the close method is used.

Eric Moreau has a post on using MARS.  It is only supported in the SQL Server and Oracle managed providers. 

You can encrypt/decrypt the data passed from the client back and to the database server by adding the Encrypt keyword to the connection string for SQL Server or using the Encrypt property on the SqlConnectionStringBuilder.

• Work with data providers.
  May include but is not limited to: limitations, behaviors, performance, installation issues, deployment issues; ODBC, Microsoft OLE DB, SqlClient, managed providers, third-party providers, native providers

I guess this is a nice way of saying, "Know everything about ADO.NET".  Some of the things I'm aware of are the just because you have a managed provider available, you still need the native dll libraries.  Oracle, for example.  The client needs the oracle client libraries in order to connect to Oracle.  I mentioned some of the shortcomings of ODBC above in regards to pooling.  A great source for third party data providers can be found at the mono project (Firebird, MySql, SQLLite, PostgresSQL, Mimer, Sybase..)

• Connect to a data source by using a generic data access interface.
  May include but is not limited to: System.Data.Common namespace classes

The common namespace is the abstract definition for all data providers.  You would use this when you aren't sure which data provider your application will ultimately use.  Using the DbProviderFactory you can instantiate objects for a specific data provider.   Suman Chakrabarti has a post showing how to use the DbProviderFactory class.

• Handle and diagnose database connection exceptions.
  May include but is not limited to: implementing try/catch handlers

Joydip Kanjilal has a post on handling exceptions in ADO.NET.  If you catch a SqlException, you would examine the number property to determine the root cause of the problem.

Next -> Section II Selecting and Querying Data

I'm retrieving data without effort now.  The Linq syntax is starting to sink in.  I don't have to look at the reference as much anymore.  I think I'm stuck, though.  I'm trying to send updates back and I was hoping to have one method per entity called "Save" that would take in the modified entities and use Linq magic to get it back to the database.  No such magic seems to exist.  I'm going to have to write a method per operation (Delete, Insert, Update).  What stinks is that the work required to do this is about the same as it was before using traditional methods.

This could be a case of programming in a closet and not having anyone currently present to bounce ideas off of.  I'm going to bounce this off some of the developers when I go back to work tomorrow.

So far, Linq has a nice intuitive syntax and allows a standardized way of querying collections.  It's also nice that it supports lazy loading and relationships natively. However, from a productivity point of view, I'm not sure it adds a lot to what we currently have. 

After getting the images downloaded and setup, I've started to develop.  After getting over a few minor bumps, I'm underway.  I've used SqlMetal to generate entity classes from the DB Schema.  Initially, the classes weren't marked as datacontracts.  This is because I didn't use the /Serialization:Unidirectional switch.  After I regenerated the entity classes, I'm in business.  My WPF application binds to the results of a WCF call that retrieves data using Linq to SQL.

I haven't been very good lately about taking time to learn the new technologies, so I'm taking today off to learn Visual Studio 2008.  I've found over the years the best way to learn a new technology is to create a demo application.  I'm going to be creating a restaurant reservation / order system, using a model proposed by Barry Williams at Database Answers.  I hope to post progress as things go along.

I'm really hoping to get into Linq, so I'm going to architect the system using the traditional 3 tiers, DB, Mid Tier and Presentation.  The Mid Tier should contain all the linq stuff, exposing the objects via a WCF Service.  I'll start with a smart client presentation, and perhaps add a web facade later today. 

I'm still waiting for the download to finish; probably another 2 hours.  So right now I'm going to read all I can about C# 3.0 and Linq.  An article I found very informative is the Linq to SQL overview at MSDN.  The nagging issue I'm trying to resolve is, "How does Linq make life easier for the Mid Tier developer"?  It looks real easy if you don't mind exposing your data model to service consumers. But if you want to abstract the data model a bit, it looks like end up writing some code because you lose a lot of convenience features.