For one of our client, I needed to setup mechanics such that admin users should be able to initiate a process in Windows Service using GDI based .NET application. We had also developed the windows service in .NET. As both of them are .NET applications, there are more than one ways to IPC between them. As in this case there was only need to give signal to the service, I choose to do it through Windows Event. .NET BCL provides wrapper around Win32 functions to do this through type named EventWaitHandle.

I completed the code like I used to do in my old days, and tried the first use of it. On call to OpenExisting, client application failed with exception “No handle of the given name exists”. My first thought was that this could be because of UAC, but in that case it should say something like access violation. I tried running application in Windows 2003 Server and it worked perfectly. I decided to check with Google and documentation. I found this, which gave me a relieving thought – “I’m not the only one in this world!” .

After some time I landed to msdn page for namespaces of kernel objects, which indicates that:

For processes started under a client session, the system uses the session namespace by default. However, these processes can use the global namespace by prepending the “Global\” prefix to the object name

changed my client code to have name prefixed with “Global\”, and everything started behaving the way I wanted it to in Windows 7 as well.

public void Operate(InputType inputObject)
{
    if (null != inputObject)
    {
        . . .
    }
} 

public void Operate(InputType inputObject)
{
    if (null == inputObject)
    {
        . . .
    }
} 

Can be replaced by

public void Operate(InputType inputObject)
{
    if (inputObject.IsNotNull())
    {
        . . .
    }
} 

public void Operate(InputType inputObject)
{
    if (inputObject.IsNull())
    {
        . . .
    }
} 

Counter argument: Some user uses email client that are not capable to show HTML messages. Or someone is doing it for security purpose. Okay, I agree, then send both the views; text view as well as HTML view. Using .NET, one can send multi view email messages, depending on email client settings and/or capability, client is expected to pick the right view and shows that to the user.

Next question is, say we have decided to send email in html and text view, can we send images and style sheets embedded with email? Answers is yes. Here is how one can send multi-part email message using .net libraries that has embedded image.

Here goes the first step, create MailMessage object and setup that with appropriate values

   1: var from = new MailAddress("admin@objectpattern.com", "Himanshu");

   2: var to = new MailAddress("himanshu@objectpattern.net");

   3: var message = new MailMessage(from, to) { Subject = "Hi from ObjectPattern"};

next step changes. As we want to have two views – text and html, we will have to create email body differently.

in these lines of code, there are certain important points

   1: //creating text view

   2: var textViewContent = "Test content for text view";

   3: var textView = AlternateView.CreateAlternateViewFromString(textViewContent, Encoding.UTF8,

   4:                                                            MediaTypeNames.Text.Plain);

   5: //creating html view 

   6: var htmlViewContent = "

Html content containing image logo

Thanks,
ObjectPattern
";

   7: var htmlView = AlternateView.CreateAlternateViewFromString(htmlViewContent, Encoding.UTF8,

   8:                                                            MediaTypeNames.Text.Html);

that I would like to bring in to your notice.

• Notice that we are not assigning Body property of MailMessage type
• While creating view, we are specifying media type name as either plain text or html text, it can also be rich text.
• We have created an img tag that referring to source as ‘cid:logo.png’

okay now how we will embed image into email message? That’s easy as well:

   1: var logo = new LinkedResource("images/logo.png", MediaTypeNames.Image.Jpeg) {ContentId = "logo.png"};

   2: htmlView.LinkedResources.Add(logo);

here in first line, first parameter of constructor (”image/logo.png” text) specifies the path of the image file. One can also supply image content as stream if its not coming directly from the file. Now remember, in above section, we had specified the src attribute of img tag, that should match with content id. Else image will not be shown correctly in email client. Also notice that we are adding this image to html view.

We are almost done now. All we need to do now is to add views to message object and sending using SMTP client.

   1: message.AlternateViews.Add(textView);

   2: message.AlternateViews.Add(htmlView);

   3:

After creating setup and deployment project, double click on dependency of .net framework in solution explorer:

Now, you should be able to see launch condition for .net framework in “Launch Conditions” tab. Until you double click for the first time, node do not appear in launch conditions. And that’s where I wasted some hours:

Select the node and see the property window, where you will find the answer of the question:

Moving back to original point, we wanted to log entry and exit of each functions. There can be more then one way to do that, but we have selected to change each functions. And to change each functions I have selected to use developer named “VS.NET macro”.

From macro it is possible to navigate through solution and projects. So macro first needs to find out all class types from a project, and then for each class, iterate through its member and find out functions that has body. If function has body, inject the code that logs line on function entry and exit.

It was easy to find function entry, but there can be many way function ends, like exception is being thrown, or based of some condition it may execute “return”, or body of the function is ended and hence return. To accommodate such cases, we decide to use “using statement” of C#.

All right, enough of the background, have look at Macro code here

To me, it make more sense to do paging of records being fetched in server memory along with view paging. What if table in question has more then 100,000 (1 lack) records?

Digging a bit further, I found a resource which explains how to do paging of records being sent from stored procedure. But it was expecting to have auto identity as primary key in the table. And we were so fortunate that we couldn’t have db design such that we can afford to have auto identity primary key, and so that solution to the paging problem. On doing some more finding, we all agreed to a solution that is described below. It’s not fully optimized, as it do not have any optimization for SQL Server to prepare result sets. But worked for our need in specific project as even in 100,000 records it was working very fast.

In our solution, stored procedure that will fetch records from table and return to caller will needs to have two addition parameters, @RowIndexFrom and @RowIndexTo

Getting customers by page will look like following,

CREATE PROCEDURE [dbo].[GetCustomersPage]

(

    @RowIndexFrom int,

    @RowIndexTo int

)

AS

BEGIN

    SET NOCOUNT ON;

    WITH IndexedCustomers AS

    (

        SELECT

            ROW_NUMBER() OVER (ORDER BY Id) AS rowIndex,

            Customers.*

        FROM

            Customers

    )

    SELECT * FROM IndexedCustomers WHERE rowIndex BETWEEN @RowIndexFrom AND @RowIndexTo

    SELECT COUNT(*)

    FROM

        Customers

END

GO

The stored procedure returns two result set, first will be records for given page (defined by different between @RowIndexFrom and @RowIndexTo parameters). And the second result set will give total number of records for give query. Second result set will help preparing pages list. So, if first page needs to be retrieved of page size 10, parameter values should be 1 and 10 respectively.

.NET is one more layer above OS, e.g. Windows. If someone needs to emulate keyboard strokes to another application from .NET there is a choice of using .NET class library. To emulate keyboard events from .NET it’s easier to use

• System.Windows.Forms.SendKeys.Send(string) and
• System.Windows.Forms.SendKeys.SendWait(string)

Recently, I used this to auto log into a application that I’m developing if it’s debug build. Just to get a change from AutoHotKey.

The problem of this is it sends keystrokes to only active application.

I have created two macros that are very common in their need. While programming, we need to create string const on regular bases. It becomes inconvenient to go to top of the class while writing a method body for same class and create constant. Macro in library attached with this post becomes handy at that time.

Same use case hold very much true for creating string resources while writing method body. And it becomes really very easy if string resource can be created while writing method body.

For those who thinks like me (or want to ), I have attached two macros with this post. Macro works for VS.NET 2005, haven't tested in any other version. But should work fine in VS.NET 2003 or VS.NET 2008