How do I structure a C# console application to efficiently use IDisposable database resources?

Question

Here's my proposed (very simplified to illustrate the problem space) design for a C# console application. The database connections implement IDisposable, and this solution doesn't allow for using the database connection objects. Can someone propose a more correct structure for a console application? This is a problem I need to solve often.

class Program 
{
    SQLiteConnection sourceConnection;
    SQLiteConnection destinationConnection;

    static void Main(string[] args)
    {
        Program shell = new Program();

        // get connection strings from command line arguments
        string sourceConnectionString = shell.getConnectionString(args);
        string destinationConnectionString = shell.getConnectionString(args);

        // call non-static methods that use
        shell.setUpConnections(sourceConnectionString, destinationConnectionString);

        shell.doDatabaseWork();
    }

    private void setUpConnections(string sourceConnectionString, string destinationConnectionString)
    {
        sourceConnection = new SQLiteConnection(sourceConnectionString);
        destinationConnection = new SQLiteConnection(destinationConnectionString);
    }

    private void doDatabaseWork()
    {
        // use the connections here
    }
}

Edit:

Some people can't figure out why I'd want them as member variables. Here's my use case (a little psuedocoded) of what would go in doDatabaseWork:

foreach (Row sourceRow in DBResultSet)
{
  string sourceXml = sourceRow.Columns["MyColumnName"].Value;
  string destinationXML = transformUsingXSLT(sourceXml);
  writeToDestination(destinationXml);
}

See how I'd want to keep these connections open for the life of this loop?

Answer 1

I think that the best solution is to extract main logic from Program class. The Program class is some kind of starter for primary work. And providing wrappers for SqlConnections is not a good idea indeed, because they are managed resources already, wrapping them is redundant. Thus my solution looks like this:

class ProgramCore : IDisposable
{
    internal ProgramCore(string sourceConnectionString, string destinationConnectionString)
    {
        setUpConnections(sourceConnectionString, destinationConnectionString);
    }

    internal void Execute()
    {
        // do whatever you want
        doDatabaseWork();
        // do whatever you want
    }

    public void Dispose()
    {
        if (_sourceConnection != null)
            _sourceConnection.Dispose();
        if (_destinationConnection != null)
            _destinationConnection.Dispose();
    }

    private void setUpConnections(string sourceConnectionString, string destinationConnectionString)
    {
        _sourceConnection = new SQLiteConnection(sourceConnectionString);
        _destinationConnection = new SQLiteConnection(destinationConnectionString);
    }

    private void doDatabaseWork()
    {
        // use the connections here
    }

    private SQLiteConnection _sourceConnection;
    private SQLiteConnection _destinationConnection;
}

class Program
{
    static void Main(string[] args)
    {
        // get connection strings from command line arguments
        string sourceConnectionString = GetConnectionString(args);
        string destinationConnectionString = GetConnectionString(args);

        using (ProgramCore core = new ProgramCore(sourceConnectionString, destinationConnectionString))
        {
            core.Execute();
        }
    }

    static string GetConnectionString(string[] args)
    {
        // provide parsing here
    }
}

Answer 2

How about writing a class that implements IDisposable.

Inside your class constructor, you can instantiate your DB connections.

Then inside your IDisposable.Dispose Method, you write your tear down code for closing your DB connections.

Here is a code sample to demonstrate what I mean:

public class DBWrapper : IDisposable
{
    public SqlConnection Connection1 { get; set; }
    public SqlConnection Connection2 { get; set; }

    public DBWrapper()
    {
        Connection1 = new SqlConnection();
        Connection1.Open();
        Connection2 = new SqlConnection();
        Connection2.Open();
    }
    public void DoWork()
    {
        // Make your DB Calls here
    }

    public void Dispose()
    {
        if (Connection1 != null)
        {
            Connection1.Dispose();
        }
        if (Connection2 != null)
        {
            Connection2.Dispose();
        }
    }
}

And then, from within your main method of your Program class:

class Program
{
    static void Main(string[] args)
    {
        using (DBWrapper wrapper = new DBWrapper())
        {
            wrapper.DoWork();
        }
    }
}

Answer 3

Scott's answer is one way to do it. You could also consider using try{} finally instead?

static void Main(string[] args)
{
    Program shell = new Program();

    // get connection strings from command line arguments
    string sourceConnectionString = shell.getConnectionString(args);
    string destinationConnectionString = shell.getConnectionString(args);

    // call non-static methods that use
    shell.setUpConnections(sourceConnectionString, destinationConnectionString);
    try
    {
      shell.doDatabaseWork();
    }
    finally
    {
      if(sourceConnection != null)
        sourceConnection.Dispose();
      if(destinationConnection != null)
        destinationConnection.Dispose();
    }
}

Answer 4

Personally, I think you are over thinking this and the code samples in this thread are overly complex imho. I have no idea why people are implementing IDisposable on their Program class either since it's disposed when it exits.

I can't think of a single reason to not use or why you cannot use the using(){} statement.

You want to open a Connection and hold it? Why? All the real connections are behind the scenes in .net connection pooling, so new'ing Connection objects is not a big deal. Just open and close as you need them and connection pooling handles all that behind the scenes.

I edited my example to wrap it in a class so you can have your encapsulation as well.

class Program 
{
    static void Main(string[] args)
    {
        DBWorker worker = new DBWorker();
        worker.DoDatabaseWork();
    }
}

public class DBWorker 
{

    private void DoDatabaseWork()
    {
        using (SQLiteConnection sourceDB = new SQLiteConnection( GetConnectionString() ))
        {
            sourceDB.Open();
            using (SQLiteConnection destDB = new SQLiteConnection( GetConnectionString() ))
            {
                destDB.Open();
            }
        }
    }

}

Answer 5

Hmm, I see no one has mentioned doing it this way. You don't have to have the variables that are used in the using declared locally.

class Program 
{
    SQLiteConnection sourceConnection;
    SQLiteConnection destinationConnection;

    static void Main(string[] args)
    {
        Program shell = new Program();

        // get connection strings from command line arguments
        string sourceConnectionString = shell.getConnectionString(args);
        string destinationConnectionString = shell.getConnectionString(args);

        using (sourceConnection = new SQLiteConnection(sourceConnectionString))
        using (destinationConnection = new SQLiteConnection(destinationConnectionString))
        {
            shell.doDatabaseWork();
        }
    }

    private void doDatabaseWork()
    {
        // use the connections here
    }
}