Raj Kaimal

VS 2010 Beta 2 Installation instructions

rajbk — Tue, 20 Oct 2009 01:39:00 GMT

For those of you that have VS 2010 Beta 1 installed.

http://go.microsoft.com/fwlink/?LinkID=166199

http://go.microsoft.com/fwlink/?LinkID=167718 (addendum)

Other observations:

I was prompted for my VS 2010 Beta 1 media when uninstalling it so have your iso or media ready.

VS 2010 Beta 2 prompted for a OS restart during installation.

See announcement for more details.

XMLSerializer and invalid XML

rajbk — Mon, 05 Oct 2009 01:29:08 GMT

A user had pasted some text from powerpoint into a textarea in one of our web apps which was eventually serialized into XML. Upon trying to de-serialize the XML, the web server threw the exception below:

System.InvalidOperationException: There is an error in XML document (1, 50). ---> System.Xml.XmlException: '♂', hexadecimal value 0x0B, is an invalid character. Line 1, position 50.

The XML that was about to be de-serialized looked like something this:

<?xml version="1.0" encoding="utf-16"?>
<string>Quick&#xB;Brown
Fox
</string>

The code used to de-serialze the xml is shown below:

    public static T DeSerializeObject<T>(string xml)

    {

        using (System.IO.StringReader sr = new System.IO.StringReader(xml))

        {

            XmlSerializer serializer = new XmlSerializer(typeof(T));

            return (T)serializer.Deserialize(sr);

        }

    }

As seen from the exception above, the de-serializer was complaining about the invalid character - .

The PowerPoint slide that text was being pasted from looked something like this:

The user had used the Shift-Enter key combination to force a line break between text in a bullet. So “Brown” is started on a new line for the same bullet as “Quick”.

By selecting all the text on the slide and pasting it into a hex editor, we see the standard hex values for carriage return and line feed – 0x0D 0x0A between “Brown” and “Fox”. But we see 0x0B was generated for the new line between “Quick” and “Brown”. 0x0B in ASCII stands for a vertical line tab which is an invalid character in XML.

The code used to serialize the string is shown below:

    private static string SerializeObject<T>(T source)

    {

        var serializer = new XmlSerializer(typeof(T));

        using (var sw = new System.IO.StringWriter())

        using (var writer = new XmlTextWriter(sw))

        {

            serializer.Serialize(writer, source);

            return sw.ToString();

        }

    }

The problem with this code is that the XmlTextWriter class which inherits from XmlWriter, does not, on its own, validate each character before serializing it.

The recommended way to serialize is to use the static XmlWriter.Create method. The method has overloads where you specify an XmlWriterSettings class. If you do not specify one, the default values for the XmlWriterSettings class is used. One of the properties of this class is the XmlWriterSettings.CheckCharacters Property which is set to true by default. This property ensures that the XmlWriter instance created by the .Create method will perform character checking. By implementing the .Create* method in our code, we ensure that the serializer will throw an exception if it encounters invalid XML.

We, therefore, can rewrite our serializer class like so:

    private static string SerializeObject<T>(T source)

    {

        var serializer = new XmlSerializer(typeof(T));

        using (var sw = new System.IO.StringWriter())

        using (var writer = XmlWriter.Create(sw))

        {

            serializer.Serialize(writer, source);

            return sw.ToString();

        }

    }

* The writer created in this case is of type System.Xml.XmlWellFormedWriter. It will vary based on the .Create overload used.

LINQ to SQL Paging Gotcha

rajbk — Wed, 30 Sep 2009 11:53:00 GMT

Framework Version: 3.5

I ran into an issue with how LINQ to SQL implements the Skip and Take operators which leads to records being repeated or missing when performing custom paging.

I am using the Northwind database. My UI will display the City and ContactName of all Customers five records at a time. The LINQ expression projects to an anonymous type with properties of City and ContactName. We see below the LINQ expression, the SQL generated and the data returned for the first and next 5 records.

First 5 Records (Page 1)		Next 5 Records (Page 2)
LINQ Expression (from c in Customers select new { c.City, c.ContactName }).Skip(0).Take(5)		LINQ Expression (from c in Customers select new { c.City, c.ContactName }).Skip(5).Take(5)
Generated SQL SELECT TOP (5) [t0].[City], [t0].[ContactName] FROM [Customers] AS [t0]		Generated SQL DECLARE @p0 Int = 5 DECLARE @p1 Int = 5 SELECT [t1].[City], [t1].[ContactName] FROM ( SELECT ROW_NUMBER() OVER (ORDER BY [t0].[City], [t0].[ContactName]) AS [ROW_NUMBER], [t0].[City], [t0].[ContactName] FROM [Customers] AS [t0] ) AS [t1] WHERE [t1].[ROW_NUMBER] BETWEEN @p0 + 1 AND @p0 + @p1 ORDER BY [t1].[ROW_NUMBER]
Result set City ContactName --------------- ------------------------------ Berlin Maria Anders México D.F. Ana Trujillo México D.F. Antonio Moreno London Thomas Hardy Luleå Christina Berglund		Result set City ContactName --------------- ------------------------------ Barquisimeto Carlos González Bergamo Giovanni Rovelli Berlin Maria Anders Bern Yang Wang Boise Jose Pavarotti

From the result set, we see that Maria Anders is repeated in Page 1 and Page 2! The same behavior can also be observed with a LinqDataSource as defined below:

<asp:GridView ID="GridView1" runat="server" AllowPaging="True" AutoGenerateColumns="False"    DataSourceID="LinqDataSource1" PageSize="5">    <Columns>        <asp:BoundField DataField="City" HeaderText="City" ReadOnly="True" SortExpression="City" />        <asp:BoundField DataField="ContactName" HeaderText="ContactName" ReadOnly="True"            SortExpression="ContactName" />    </Columns></asp:GridView><asp:LinqDataSource ID="LinqDataSource1" runat="server" ContextTypeName="DataClassesDataContext"     Select="new (City, ContactName)" TableName="Customers"></asp:LinqDataSource>

The reason for this behavior is because the SQL generated for Page 1 does not have an order specified which results in the records being returned in an unknown order

SELECT TOP (5) [t0].[City], [t0].[ContactName]
FROM [Customers] AS [t0]

while for subsequent pages, the SQL generated has an order by clause (required for the ROW_NUMBER() function).

(ORDER BY [t0].[City], [t0].[ContactName])

While paging through the result set, this will either result in repeating records or missing records*.

The best way to work around this issue in framework version 3.5 is to always specify an orderby in the expression before the Take and Skip operators. By doing this, you guarantee that the order used for the first page (with the TOP operator) will be the same as the order used for subsequent pages (using the ROW_NUMBER() OVER ORDER function)

(from c in Customers orderby c.ContactName        select new {            c.City,            c.ContactName        }).Skip(0).Take(5)

SELECT TOP (5) [t0].[City], [t0].[ContactName]FROM [Customers] AS [t0]ORDER BY [t0].[ContactName]

LINQ To Entities on the other hand forces you to specify an order by in a scenario like this. If you don’t you get the following error:

The method 'Skip' is only supported for sorted input in LINQ to Entities.
 The method 'OrderBy' must be called before the method 'Skip'.

This issue seems to have been fixed in the .NET 4.0 framework where the SELECT TOP statement is no longer generated by LINQ TO SQL for the first page.

(from c in Customers orderby c.ContactName        select new {            c.City,            c.ContactName        }).Skip(0).Take(5)

DECLARE @p0 Int = 0DECLARE @p1 Int = 5 SELECT [t1].[City], [t1].[ContactName]FROM (    SELECT ROW_NUMBER() OVER (ORDER BY [t0].[ContactName]) AS [ROW_NUMBER], [t0].[City], [t0].[ContactName]    FROM [Customers] AS [t0]    ) AS [t1]WHERE [t1].[ROW_NUMBER] BETWEEN @p0 + 1 AND @p0 + @p1ORDER BY [t1].[ROW_NUMBER]

In addition, if you don’t specify an orderby clause or project your result, LINQ To SQL will use all the properties specified in the entity class in the order by clause and select statement leading to ugly and inefficient code:

(from c in Customers        select c).Skip(10).Take(5)

SELECT [t1].[CustomerID], [t1].[CompanyName], [t1].[ContactName], [t1].[ContactTitle], [t1].[Address], [t1].[City], [t1].[Region], [t1].[PostalCode], [t1].[Country], [t1].[Phone], [t1].[Fax]FROM (    SELECT ROW_NUMBER() OVER (ORDER BY [t0].[CustomerID], [t0].[CompanyName], [t0].[ContactName], [t0].[ContactTitle], [t0].[Address], [t0].[City], [t0].[Region], [t0].[PostalCode], [t0].[Country], [t0].[Phone], [t0].[Fax]) AS [ROW_NUMBER], [t0].[CustomerID], [t0].[CompanyName], [t0].[ContactName], [t0].[ContactTitle], [t0].[Address], [t0].[City], [t0].[Region], [t0].[PostalCode], [t0].[Country], [t0].[Phone], [t0].[Fax]    FROM [Customers] AS [t0]    ) AS [t1]WHERE [t1].[ROW_NUMBER] BETWEEN @p0 + 1 AND @p0 + @p1ORDER BY [t1].[ROW_NUMBER]

Special thanks to Joseph Albahari for creating LINQPad.

* In most cases we get lucky and get the records from SQL based on the PK for the first page (even though as mentioned before, order is not guaranteed) and since our LINQ Expression will most likely have the PK as the first item in the projection it results in the same order for both the first page and subsequent pages. Hence this obscure issue.

Windows 7 Backup

rajbk — Tue, 29 Sep 2009 17:18:50 GMT

The backup tool that comes with Vista does not allow users to select individual folders for backup. Instead the user is presented with a default list of locations for backup. Because of this limitation, I had to resort to other third party tools to do the job.

With Windows 7, you now have the option of selecting individual folders for backup. Nice!

Windows 7 – Start Bar and Task Bar

rajbk — Sat, 08 Aug 2009 17:42:46 GMT

I Installed 7 RTM yesterday and I am really impressed by the speed improvements compared to Windows Vista.

Even though I am still learning the new features, there are a couple of things I have seen that I really like so far.

You can now click on the network icon in the taskbar and easily connect or disconnect to any network you have configured including VPN connections.

I also like how easy it is to get to your remote desktop connections. You can click on the start/windows icon and get a list of the recent connections made and even pin down your favorites.

You can even pin down your favorite programs to the taskbar.

You can also easily control which icons appear in the notification area and reduce clutter. For example you can turn off the ASP.net web server notification by clicking on the arrow in the taskbar and selecting customize :

Check out this post for lots more features

Looking awesome so far!

FormView Binding Gotcha

rajbk — Mon, 03 Aug 2009 06:17:00 GMT

Version: ASP.NET 3.5 SP1

When performing two way databinding, Null values and Nullables get changed to string.Empty in the FormView.
The way the FormView control handles null values has caused me some grief :-(

Consider a FormView bound to an ObjectDataSource like so:

<asp:FormView ID="FormView1" runat="server" DataSourceID="ObjectDataSource1" DefaultMode="Edit">
    <EditItemTemplate>
        FirstName:
        <asp:TextBox ID="FirstNameTextBox" runat="server" Text='<%# Bind("FirstName") %>' />
        <br />
        LastName:
        <asp:TextBox ID="LastNameTextBox" runat="server" Text='<%# Bind("LastName") %>' />
        <br />
        <asp:LinkButton ID="UpdateButton" runat="server" CausesValidation="True" CommandName="Update"
            Text="Update" />
    </EditItemTemplate>
</asp:FormView>

<asp:ObjectDataSource ID="ObjectDataSource1" runat="server" DataObjectTypeName="Employee"
        OldValuesParameterFormatString="original_{0}" SelectMethod="GetEmployee" TypeName="EmployeeWrapper"
        ConflictDetection="CompareAllValues" UpdateMethod="SetEmployee"></asp:ObjectDataSource>

The Employee object returned by the ObjectDataSource has LastName set to null like so:

[DataObjectMethod(DataObjectMethodType.Select)]
public Employee GetEmployee()
{
    return new Employee()
    {
        FirstName = "Tom",
        LastName = null
    };
}

Click the Update button on the FormView without changing any properties. In the Form ItemUpdating event, we see that both the OldValues and NewValues have the LastName property set to string.Empty.

(OldValue according to FormView)
FirstName : Tom
LastName :
 
(NewValue according to FormView)
FirstName : Tom
LastName :

On the other hand, if we try the same thing with the DetailsView control, the LastName will still be “null” on update.

This FormView behavior can be problematic when we we need the original object and the modified object to perform an update with conflict detection (see LINQ to SQL example below). Note that ConflictDetection is set to CompareAllValues in the ObjectDataSource.

<asp:ObjectDataSource ID="ObjectDataSource1" runat="server" ConflictDetection="CompareAllValues"
    DataObjectTypeName="Customer" OldValuesParameterFormatString="original_{0}" SelectMethod="GetCustomer"
    TypeName="CustomerWrapper" UpdateMethod="SetCustomer">
    <UpdateParameters>
        <asp:Parameter Name="customer" Type="Object" />
        <asp:Parameter Name="original_customer" Type="Object" />
    </UpdateParameters>
</asp:ObjectDataSource>

public void SetCustomer(Customer customer, Customer original_customer)
{
    DataClassesDataContext context = new DataClassesDataContext();
    context.Customers.Attach(customer, original_customer);
    context.SubmitChanges();
}

In brief, LINQ to SQL compares the original value and new value and generates SQL code to update the original object. Since the original object was modified by the FormView, it will never be found by SQL and no update will occur!

While there are other techniques you can use with Linq to SQL to avoid the issue above, the point of this post is to show how the FormView handles null values.

The workaround is to subscribe to the FormView ItemUpdating event and look for string.empty NewValues and set those to null instead.

protected void FormView2_ItemUpdating(object sender, FormViewUpdateEventArgs e)
{
    if (e.NewValues["LastName"].Equals(string.Empty))
    {
        e.NewValues["LastName"] = null;

    }
}

In addition, to keep the OldValues unchanged (if you need it for concurrency purposes), set the DataKeyNames property of the FormView to the properties of the object like so:

<asp:FormView ID="FormView2" runat="server" DataSourceID="ObjectDataSource1" OnItemUpdating="FormView2_ItemUpdating"
        DefaultMode="Edit" DataKeyNames="LastName, FirstName">

When the Update button is clicked this time, the null values are retained.

(OldValue according to FormView)
FirstName : Tom
LastName : null
 
(NewValue according to FormView)
FirstName : Tom
LastName : null

Two way binding is not supported for nested controls – AFAIK, this is by design.
See an example below:

<EditItemTemplate>
    <asp:UpdatePanel runat="server" ID="up1">
        <ContentTemplate>
            FirstName:
            <asp:TextBox ID="FirstNameTextBox" runat="server" Text='<%# Bind("FirstName") %>' />
        </ContentTemplate>
    </asp:UpdatePanel>
    LastName:
    <asp:TextBox ID="LastNameTextBox" runat="server" Text='<%# Bind("LastName") %>' />
    <br />
    <asp:LinkButton ID="UpdateButton" runat="server" CausesValidation="True" CommandName="Update"
        Text="Update" />
</EditItemTemplate>

Here we have a TextBox inside an UpdatePanel inside the EditItemTemplate. Since two way binding is not supported for the FirstNameTextBox, the special code for extracting values is not generated resulting in null being sent back for the FirstName property.

To work around this, you will need to subscribe to the ItemUpdating event and extract the values yourself:

protected void FormView2_ItemUpdating(object sender, FormViewUpdateEventArgs e)
{
    TextBox FirstNameTextBox = (TextBox)FormView2.FindControl("FirstNameTextBox");
    e.NewValues["FirstName"] = FirstNameTextBox.Text;
}

Visibility of controls in a FormView.
Setting Visibility of a control inside a FormView does not work - the TextBox control goes back to visible on update.

protected void Page_Load(object sender, EventArgs e)
{
    FormView1.FindControl("FirstNameTextBox").Visible = false;
}

Do it in the DataBound event instead:

protected void FormView2_DataBound(object sender, EventArgs e)
{
    FormView1.FindControl("FirstNameTextBox").Visible = false;
}

If you have better ideas on any of the workarounds above, please post a comment or link. Thanks.

Altering Indexed Views - Gotcha

rajbk — Thu, 16 Jul 2009 16:25:00 GMT

According to BOL, “ALTER VIEW can be applied to indexed views; however, ALTER VIEW unconditionally drops all indexes on the view.”

We can see this behavior by creating an indexed view using the Northwind database. The script is shown below:

CREATE VIEW [dbo].[MyView] WITH SCHEMABINDING
AS
SELECT     ContactName, CompanyName
FROM         dbo.Customers

GO

CREATE UNIQUE CLUSTERED INDEX [inx_MyIndex] ON [dbo].[MyView] 
(
    [ContactName] ASC
)
GO

Note that the indexed view must be created with the SCHEMABINDING option. SCHEMABINDING binds the view to the schema of the underlying base tables.

Now the fun part. We alter the index view by adding a new column “ContactTitle”, like so:

ALTER VIEW [dbo].[MyView] WITH SCHEMABINDING
AS
SELECT     ContactName, CompanyName, ContactTitle
FROM         dbo.Customers

GO

-- Output from Management Studio
-- Command(s) completed successfully.

By looking at the view in management studio, after refreshing the tree, we see that the index is no longer there! I expected to see a warning message when this occurred :-(

The easiest way to work around this is to make a copy of the index creation script before you alter the indexed view and run it afterwards.

Using the Modal Popup Extender to build a popup search interface

rajbk — Thu, 07 May 2009 06:30:05 GMT

This post and sample code demonstrates how to use the Modal Popup Extender (MPE) to display a popup search box, select a record from the popup, hide the popup and display details for the selected record on the page using AJAX.

We will be using the Northwind database and displaying a “Find Customer” popup. Once a Customer is selected from the search result list, we hide the MPE and refresh UpdatePanels on the page with information related to the Customer that was picked. Sample source code is attached at the bottom of the page.

The events that occur as as follows:

When we click on the “Show Customer Picker” button, we have a popup appear with the help of an MPE. This MP has a UserControl with textboxes, a search button and a GridView inside an UpdatePanel. Performing a “Search” will cause only the contents of this UpdatePanel to get refreshed. To avoid the search GridView from binding when the popup is hidden (when an async postback occurs by some other button on the page), we keep track of the MP visibility (The technique for keeping track of the MPE visibility is described here). If the MP is hidden, we use the ObjectDataSourceSelectingEventArgs.Cancel method in the ObjectDataSource Selecting event to cancel the SQL call.

When we select an employee by clicking on the “Select” link, the Selecting event of the GridView is raised. We get the primary key of the selected customer by subscribing to this event. The MP is then hidden and a custom CustomerSelected event is raised.

The Page is subscribed to the CustomerSelected event and saves the selected customer primary key to Session (You could use other techniques instead of this). It then forces the UpdatePanels of other UserControls on the page to refresh themselves. These UpdatePanels have databound controls in them that get bound by making SQL calls using the customer PK stored in Session. The page ends up showing information about the selected customer (Customer Details, Last 10 orders, To 10 Orders) on the page without a full page refresh.

The sample website project shows you how to achieve this UI in three stages (three pages). The demo in Stage1 shows how the page is laid out without using AJAX – all postbacks result in a full page refresh. In Stage2, we add the UpdatePanels to perform partial page rendering. In Stage3, we add the MP that displays the “Find Customer” popup.

Don’t forget to add a reference to the Ajax Control ToolKit before building the project.

Source Code
Mirror

Check Modal Popup Extender visibility from code behind

rajbk — Tue, 05 May 2009 05:41:50 GMT

AFAIK, the Modal Popup Extender has no direct way to determine its visibility state from code behind. This post describes a workaround for that.

The idea here is to wire up handlers that the MPE fires just before it’s about to show the popup (showing event) and before it’s about to hide the popup (hiding event).

In the handlers, we set the property of a HiddenField webcontrol to either ‘1’ or ‘’. We then check the Value property of this HiddenField from code behind. If the MPE has a value of 1, we know the MPE is visible.

   1: Sys.Application.add_load(applicationLoadHandler);

2:

3:

   4: //Subscribe to the show and hide events of the modal popup.

   5: //Set a hidden field some value when visible and set to empty when hidden

   6: //This hidden field is used in code behind to determine the popup visibility.

   7: function applicationLoadHandler() {

   8:     var mpeEmployeeSearch = $find('mpeEmployeeSearch');

   9:     if (mpeEmployeeSearch) {

  10:         mpeEmployeeSearch.add_showing(employeeShowingHandler);

  11:         mpeEmployeeSearch.add_hiding(employeeHidingHandler);

  12:     }

  13: }

14:

  15: function employeeShowingHandler() {

  16:     $get('hfModalVisible').value = '1';

  17: }

18:

  19: function employeeHidingHandler() {

  20:     $get('hfModalVisible').value = '';

  21: }

The MPE in this case has been assigned a BehaviorID of mpeEmployeeSearch. The HiddenField WebControl has ID (rendered in HTML) of “hfModalVisible”.

Finally, we can check the MPE visibility from code behind like so:

   1: if (!string.IsNullOrEmpty(hfModalVisible.Value))

   2: {

   3:     //Do something

   4: }

Post a comment if you know of a better way.

Uploading an Excel file to SQL through an ASP.NET webform

rajbk — Sat, 02 May 2009 21:59:00 GMT

The method below describes how to upload a file to a webserver and then import the file into SQL using either LinqToSQL or SQL Bulk Copy. The sample code only shows how to import xls and xlsx files but it could easily be extended to support csv files too. Sample code is attached at the bottom.

We will be uploading data from an Excel file containing columns CompanyName and Phone and loading that into the Northwind Shippers table.

We’ll start by uploading the file to the webserver. This is done with the help of the FileUpload web control. The FileUpload control has a SaveAs method which saves the contents of the file into the location that we specify. The file will be stored in a temp folder under App_Data since App_Data is not browsable directly by users.

Once we have successfully uploaded the file to the webserver, we use an OleDbConnection and an OleDbDataReader to read each row from the Excel file. The OleDb connection string varies by file extension. The connection strings are shown below:

Extension	ConnectionString
xls	Provider=Microsoft.Jet.OLEDB.4.0;Data Source=[FilePath];Extended Properties=”Excel 8.0;HDR=YES;IMEX=1”
xlsx	Provider=Microsoft.ACE.OLEDB.12.0;Data Source=[FilePath];Extended Properties=Excel 12.0 Xml;HDR=YES;IMEX=1

HDR=Yes specifies that the first row of the data contains column names and not data
IMEX=1 specifies that the driver should always read the “intermixed” data columns as text.

The query we will be using with the connection is "SELECT CompanyName, Phone FROM [Sheet1$]". This assumes that we have an excel sheet called Sheet1 with header columns CompanyName and Phone.

Method 1: Using LINQ To SQL

Using the OleDBDataReader, we read each record and create a new Shipper object for each OleDbDataReader record as shown below. We add this object to the Shipper collection object that is associated with the Shipper table in the database using InsertOnSubmit and call SubmitChanges. This loads all the Excel records into the Shipper table.

Note: Since we are calling SubmitChanges without any Transaction defined, LINQ to SQL automatically starts a local transaction and uses it to execute the insert statements. When all insert statements successfully complete, LINQ to SQL commits the local transaction – nice:-) This occurs behind the scenes.

   1: //ref: http://msdn.microsoft.com/en-us/library/system.data.oledb.oledbdatareader(VS.71).aspx   2: using (var context = new NorthwindDataContext())   3: {   4:     using (var myConnection = new OleDbConnection(base.SourceConnectionString))   5:     using (var myCommand = new OleDbCommand(query, myConnection))   6:     {   7:         myConnection.Open();   8:         var myReader = myCommand.ExecuteReader();   9:         while (myReader.Read())  10:         {  11:             context.Shippers.InsertOnSubmit(new Shipper()  12:             {  13:                 CompanyName = myReader.GetString(0),  14:                 Phone = myReader.GetString(1)  15:             });  16:         }  17:     }  18:    19:     context.SubmitChanges();  20: }

Method 2: Using SQL BulkCopy

With the BulkCopy method, we first have to define the Column Mappings since we will not be inserting data into the autogenerated ShipperID Primary Key column. The first column in the Excel file (CompanyName) has to be mapped to the second column in the Shipper table and the second column (Phone) has to be mapped to the third column in the Shipper table as shown below.

We read each record from the OleDbDataReader and using the BulkCopy WriteToServer overload that takes in an IDataReader (which the OleDbDataReader implements). The BulkCopy, using this method bulk loads the Shippers destination table with the data from the OleDbDatareader.

   1: //ref: http://msdn.microsoft.com/en-us/library/system.data.oledb.oledbdatareader(VS.71).aspx   2: //ref: http://msdn.microsoft.com/en-us/library/system.data.sqlclient.sqlbulkcopy.aspx   3: using (var myConnection = new OleDbConnection(base.SourceConnectionString))   4: using (var destinationConnection = new SqlConnection(destinationConnectionString))   5: using (var bulkCopy = new SqlBulkCopy(destinationConnection))   6: {   7:     //Map first column in source to second column in sql table (skipping the ID column).   8:     //Excel schema[CompanyName,Phone] Table schema[ShipperID, CompanyName, Phone]   9:     bulkCopy.ColumnMappings.Add(0, 1);  10:    11:     bulkCopy.ColumnMappings.Add(1, 2);  12:     bulkCopy.DestinationTableName = "dbo.Shippers";  13:    14:     using (var myCommand = new OleDbCommand(query, myConnection))  15:     {  16:         myConnection.Open();  17:         destinationConnection.Open();  18:    19:         var myReader = myCommand.ExecuteReader();  20:         while (myReader.Read())  21:         {  22:             bulkCopy.WriteToServer(myReader);  23:         }  24:     }  25: }

You will need the 2007 Office System Driver : Data Connectivity Components installed on your server also.

The BulkCopy object is much faster than LINQ to SQL. I am copying Pablo Castro’s newsgroup response:

No per-row statement execution. When you do multiple inserts without bulk-copy, each insert is a statement in itself (regardless of whether it's batched together with other statements). With bulk-copy, we don't incur the cost of executing a statement for each row, the whole copy operation is a single thing.
No multiple network round-trips. Once the bulk-insert operation is setup, we send rows from the client to the server continously, without going back-and-forth over the wire.
Server storage engine also can greatly optimize how rows are inserted when performing a bulk-copy operation. How much can be optimized depends a lot on the recovery model the tarder database is set to; in "simple" and "bulk logged" the overhead of logging is greatly reduced during bulk-copy operations, helping a lot with performance.

Sample Code
Mirror

Implementing Deeplinking in Silverlight

rajbk — Wed, 09 Jul 2008 07:58:00 GMT

This article is now outdated. Silverlight 3 now has deeplinking built in. There might be other articles on this but this is the only one I could find on this topic: http://programwith.net/2009/03/23/Silverlight3NdashDeepLinking.aspx

Version: Silverlight 2 Beta 2 / .NET 3.5 SP1

See source code links at the bottom of this post.

The following post is an attempt at implementing deep linking in Silverlight. Deeplinking allows users to use the browser back and forward buttons to switch between different “states” of the application and also allows bookmarking a certain state of the app. It provides a great UX in scenarios like viewing a slide show, paging through a document etc. IMHO, I think deeplinking should be baked into Silverlight by default.

To implement this, we use the System.Web.Extensions assembly in .NET 3.5 SP1 Beta which adds native support for navigating forward and backward through the browser history stack and also allows bookmaking. We will add some custom code to our Silverlight application and host it inside an aspx page so that it can take advantage of the new features of the assembly or more specifically the JavaScript code in this assembly.

As a quick refresher, in the context of an ASP.NET AJAX application, we add new entries to the browser history stack by calling Sys.Application.addHistoryPoint(state, title) where state is a StringDictionary containing the state we wish to store and title is the page title.

The Sys.Application.navigate event is raised anytime the user clicks the browser back or forward button. To restore the state of the page, we subscribe to this event and call e.get_state() inside our event handler. You can read more on this topic here.

The idea here is to have our Silverlight application make a javascript call to addHistoryPoint anytime we wish to store the state of the application in the browser history. We then subscribe to the JavaScript Sys.Application.navigate event and change the state of our application when this event is raised.

We start by creating Silverlight application using Blend 2.5 June 2008 preview or VS 2008. In the App.cs class, we add a helper method for making the call to addHistoryPoint like so (FYI, HtmlPage.Window.Eval method performs the equivalent of making a JavaScript eval method call):

   1: /// <summary>   2: /// Call to Sys.Application.addHistoryPoint   3: /// </summary>   4: public void AddHistoryPoint(string stateKey, string stateValue, string title)   5: {   6:     string addHistoryScript = "Sys.Application.addHistoryPoint({{ {0}:'{1}' }}, '{2}');";   7:     HtmlPage.Window.Eval(string.Format(addHistoryScript, stateKey, stateValue, title));   8: }

The next step is to subscribe to the JavaScript Sys.Application.navigate event from our Silverlight managed code. To make our managed code accessible through javascript, we have to register our class as a scriptable object and then make our method accessible by marking it with the Scriptable attribute, Read more on this topic here.

As shown below, HandleNavigate is marked as a scriptable member which exposes it to JavaScript.

   1: /// <summary>   2: /// Events   3: /// </summary>   4: public event EventHandler<StateEventArgs> Navigate;   5:     6: /// <summary>   7: /// Handler for Sys.Application navigate event   8: /// </summary>   9: [ScriptableMember]  10: public void HandleNavigate(ScriptObject state)  11: {  12:     if (Navigate != null)  13:     {  14:         Navigate(this, new StateEventArgs() { State = state });  15:     }  16: }

In the code below, we register our class as a scriptable object (Line 13). We also dynamically load a Javascript script block which creates a helper JavaScript function that subscribes to the navigate event. The helper function then makes a call to our silverlight HandleNavigate method (Line 17). The last method in the script block is a call to __navigateHandler which is for restoring the state of the page when the page loads for the first time or through a bookmark.

   1: private void OnStartup(object sender, StartupEventArgs e)   2: {   3:     // Load the main control here   4:     this.RootVisual = new Page();   5:     InitScripts();   6: }   7:     8: /// <summary>   9: /// Register event hander through proxy method  10: /// </summary>  11: private void InitScripts()  12: {  13:     HtmlPage.RegisterScriptableObject("App", this);  14:     string initScript = @"  15:         var __navigateHandler = new Function('obj','args','document.getElementById(\'appId\').content.App.HandleNavigate(args.get_state())');  16:         Sys.Application.add_navigate(__navigateHandler);  17:         __navigateHandler(this, new Sys.HistoryEventArgs(Sys.Application._state));  18:      ";  19:     HtmlPage.Window.Eval(initScript);  20: }

With this in place, we can subscribe to the Navigate event from any class and use the state information to make changes to the state of our application.

   1: App app = (App)Application.Current;   2: app.Navigate += new EventHandler<StateEventArgs>(app_Navigate);

We call the AddHistoryPoint method to store the state of the app when needed as shown below.

   1: App app = (App)Application.Current;   2: string pageTitle = (activeSlideIndex == -1) ? "Home" : "Slide " + activeSlideIndex;   3: app.AddHistoryPoint("index", activeSlideIndex.ToString(), pageTitle);

The last step involved is to create an ASP.NET AJAX application, add a ScriptManager control, set the EnableSecureHistorState to false (so that our URL is human readable) and finally add a Silverlight control with the source set to the xap file.

   1: <asp:ScriptManager ID="ScriptManager1" runat="server" EnableHistory="true" EnableSecureHistoryState="False">   2: </asp:ScriptManager>   3: <div id="slcontent">   4:     <asp:Silverlight ID="appId" runat="server" Source="~/ClientBin/SilverlightHistory.xap"   5:         MinimumVersion="2.0.30523" Width="800" Height="600" BackColor="Black" />   6: </div>

You can download the sample project here or here (This is experimental beta code - use at your own risk).
Things you can try:

Clicking on the links in the right menu adds history points to the browser stack.
You can use the browser back and forward buttons to walk the stack.
You can bookmark a URL and return to it (example: localhost:9999/SilverlightHistory_Web/Default.aspx#index=2 will load the app with the third slide visible)

Another topic that has been making headlines recently is RIA SEO. Here are three articles on that:

Comments are always welcome.

Loading an XML file in your Silverlight project into memory

rajbk — Tue, 08 Jul 2008 06:36:00 GMT

Version : Silverlight 2 Beta 2

Reading an XML file in an XAP package can easily be done with the help of a helper class like so:

   1: public static class XmlHelper   2: {   3:     public static XElement LoadDocument(string fileName)   4:     {   5:         //No longer required in Silverlight 2 since   6:         //XmlXapResolver has been added as the default resolver for XmlReader   7:         //ref: http://msdn.microsoft.com/en-us/library/cc189007(vs.95).aspx   8:         //XmlReaderSettings settings = new XmlReaderSettings();   9:         //settings.XmlResolver = new XmlXapResolver();  10:         //XmlReader reader = XmlReader.Create(fileName, settings);  11:    12:         XmlReader reader = XmlReader.Create(fileName);  13:         XElement element = XElement.Load(reader);  14:         return element;  15:     }  16: }

Once you have the XElement object, you are free to manipulate it with LINQ to XML as shown below:

   1: public static List<Slide> LoadSlides()   2: {   3:     XElement element = XmlHelper.LoadDocument("Slides.xml");   4:     var slides = from slide in element.Descendants("Slide")   5:                  select new Slide {   6:                     Title = (string)slide.Element("Title"),   7:                     Text = (string)slide.Element("Text"),   8:                     Image = (string)slide.Element("Image"),   9:                  };  10:     return slides.ToList();  11: }

The method above returns a List<Slide> where the Slide class is defined like so:

   1: public class Slide   2: {   3:     public string Title { get; set; }   4:     public string Text { get; set; }   5:     public string Image { get; set; }   6: }

and the contents of the XML file we are reading is:

   1: <?xml version="1.0" encoding="utf-8"?>   2: <Slides>   3:   <Slide>   4:     <Title>Slide 1</Title>   5:     <Text>Slide 1 Description</Text>   6:     <Image>Slide1.jpg</Image>   7:   </Slide>   8:   <Slide>   9:     <Title>Slide 2</Title>  10:     <Text>Slide 2 Description</Text>  11:     <Image>Slide2.jpg</Image>  12:   </Slide>  13: </Slides>

Note that the XML file should have its BuildAction set to "Content".

A four stroke engine in Silverlight

rajbk — Mon, 02 Jun 2008 07:26:00 GMT

I decided to try out Expression Blend 2.5 beta by creating a four stroke engine animation. A screen capture is shown below:

You can see a working demo here (Silverlight 2 plugin required)

Here are a few observations while working on this project.

Writing XAML by hand is difficult :-) Expression Blend is your friend.
To export Adobe Illustrator files to XAML, use the free plugin from Mike Swanson. I chose to use AI to create the vector art because I am familiar with AI and it has a lot more features than Expression Design. When importing the XAML into Blend, you might see error messages in Blend about invalid markup as a result of the translation. In most cases these errors can easily be fixed by editing the XAML markup (Blend gives you the line number of the error).
For performing simple custom animations using a timer in your UI thread, use the DispatcherTimer class. The timer raises a Tick event at the end of the specified interval where you can update the UI. Note that the DispatcherTimer is single threaded and does not use the thread pool which could lead to a frozen UI if the Tick event handler takes too long to complete or if the thread is busy doing something else.
To apply multiple transformations to an object, for example a RotateTransform and a ScaleTransform at the same time, use the TransformGroup class.
Expression Blend, AFAIK, does not allow you to edit your C# files. You edit them in VS 2008. Be prepared for a lot of switching back and forth between both applications. In addition, VS 2008 sometimes goes crazy and refuses to show you the XAML preview when you have a complex XAML document. Restarting VS or recompiling the project usually makes the problem go away. Probably because I am using a beta product.
Animations can be created in expression blend using a timeline where you set keyframes that define the start and end points of the transform on an object. I chose, instead, to create the animations in code behind using C# using mathematical equations.

Overall, Expression Blend 2.5 beta was easy to work with. The UI could be improved a lot though. I plan to post feedback on Microsoft Connect soon.

You can download the VS 2008 project here or from the mirror here.

Connection strings in LINQ to SQL classes.

rajbk — Wed, 21 May 2008 03:12:00 GMT

Version : VS 2008 RTW

When you have a team working on a project that contains a LINQ to SQL class (dbml), you might see the following message when trying to add a Table entity or stored procedure in a dbml created by a fellow developer:

The objects you are adding to the designer use a different data connection than the designer is currently using. Do you want to replace the connection used by the designer?

The reason this happens is because the connection string in Server Explorer used to add the new stored procedure is different from what was originally used. In a team environment, this will occur if one developer checks in the dbml using one connection string and another developer checks out the dbml and tries to add a stored procedure or a table using a different connection string. Connection strings generally fall into one of the two shown below for SQL server:

Windows Authentication:

     Data Source=.\sqlexpress;Initial Catalog=northwind;Integrated Security=True
  

SQL Authentication:

    Data Source=.\sqlexpress;Initial Catalog=Northwind;Persist Security Info=True;User ID=northwind_web;Password=**
  

The connection string could easily vary by the type of authentication, the userid/password combination, instance name etc. This could become a source control nightmare with each developer checking in the dbml file with a connection string defined in their "Server Explorer". Therefore it is best, when possible, that all developers use the same connection string defined in "Server Explorer" when working with a dbml file for a given project.

This brings up the question of how to change connection string when the dbml class library is used in an ASP.net website. The solution is create a connection string in web.config with the same name as what is specified in app.config.

To understand why, we have to look at the class library dll. I have a sample project attached at the bottom of this post that contains a class library containing a dbml and the class library being referenced from an ASP.Net project. When the library gets built, a sealed class called Settings is generated internally. Here is the class with the help of reflector:

    [CompilerGenerated, GeneratedCode("Microsoft.VisualStudio.Editors.SettingsDesigner.SettingsSingleFileGenerator", "9.0.0.0")]

    internal sealed class Settings : ApplicationSettingsBase

    {

        // Fields

        private static Settings defaultInstance = ((Settings) SettingsBase.Synchronized(new Settings()));

        // Properties

        public static Settings Default

        {

            get

            {

                return defaultInstance;

            }

        }

        [ApplicationScopedSetting, SpecialSetting(SpecialSetting.ConnectionString), DefaultSettingValue(@"Data Source=.\sqlexpress;Initial Catalog=northwind;Integrated Security=True"), DebuggerNonUserCode]

        public string northwindConnectionString

        {

            get

            {

                return (string) this["northwindConnectionString"];

            }

        }

    }

We see that the northwindConnectionString property has a DefaultSettingValue attribute containing the connection string. When Settings gets loaded at runtime, if it finds the same a connection string with name MyClassLibrary.Properties.Settings.northwindConnectionString in app.config or web.config, it will use that. If it does not find it, it reverts to the DefaultSettingValue. As long as our web.config has the same "name" as what the Settings class is looking for, that will get used in place of the default value.

What about the case where we want one connection string during development and another for production? In web.config, you can specify the configSource attribute in the connectionStrings section. This attribute specifies the fully qualified name and location of an external configuration file that contains the connectionStrings section. During development, you can have configSource point to the connectionString of your test server and when you are ready to deploy to staging or production, you can change the attribute to point to the appropriate external config file. You can also have this done automatically using Web Deployment Projects or MSBuild.

The sample project attached contains a class library with a connection string defined. This class library is used in an ASP.net website project containing a connection string in web.config with the same "name" as what was used in the class library thereby "overriding" the class library connection string value. The website project also has a configSource defined in web.config pointing to an external configuration file ConnectionStrings_Test.config. When deploying to production, the configSource will be set to point to ConnectionStrings_Prod.config which contains the production server connection string.

Comments are welcome. Thanks!

LinqDataSource exceptions

rajbk — Fri, 01 Feb 2008 02:05:00 GMT

Prerequisite: LinqDataSource & SqlDataSource Master/Details

When working with the LinqDataSource, you may get the exceptions listed below.

1. Operator '==' incompatible with operand types 'Int32' and 'Object'

The exception occurs because anytime a ControlParameter in the WhereParameters collection (IDictionary<string, object>) is null, it gets treated as type Object causing the LINQ dynamic expression parser comparison to fail. Consider the code snippet below:

     
    <asp:LinqDataSource ID="LinqDataSource2" runat="server" ContextTypeName="DataClassesDataContext"
    
        Select="new (OrderID, ProductID, UnitPrice, Quantity, Discount, Order)" TableName="Order_Details"
    
        Where="OrderID == @OrderID">
    
        <WhereParameters>
    
            <asp:ControlParameter ControlID="GridView1" Name="OrderID" PropertyName="SelectedValue"
    
                Type="Int32" />
    
        </WhereParameters>
    
    </asp:LinqDataSource>

This is part of the classic Master/Details scenario where the LinqDataSource fetches the Order Details based on the OrderID selected in the GridView1.

When the page loads the first time, the OrderID ControlParameter is equal to GridView1.SelectedValue. GridView1.SelectedValue is null since no record has been selected in the GridView1 yet. Unfortunately, the LinqDataSource still attempts to fetch the data. The Linq expression parser treats the null parameter as type Object and the comparison fails because it was expecting type Int32.

What we need here is a way to prevent the Select from occurring when any parameter is null. The LinqDataSource, unlike the SqlDataSource, for some reason, does not have a CancelSelectOnNullParameter property. This property when set to true, will cancel the select when any parameter in the SelectParameters collection is null.

We can implement this by handling the Selecting event of the LinqDataSource and call the Cancel method when any WhereParameter is null like so:

    
    protected void LinqDataSource2_Selecting(object sender, LinqDataSourceSelectEventArgs e)
    
    {
    
        foreach (KeyValuePair<string, object> kvp in e.WhereParameters)
    
        {
    
            if (kvp.Value == null)
    
            {
    
                e.Cancel = true;
    
                return;
    
            }
    
        }
    
    }

2. Operator '==' incompatible with operand types 'Guid' and 'String'

This exception occurs in cases where a parameter in the WhereParameters collection is of type Guid.

    
    <asp:LinqDataSource ID="LinqDataSource2" runat="server" 
    
        ContextTypeName="DataClassesDataContext" 
    
        Select="new (DummyID, FirstName, LastName)" TableName="DummyTables" 
    
        Where="DummyID == @DummyID" onselecting="LinqDataSource2_Selecting">
    
        <WhereParameters>
    
            <asp:ControlParameter ControlID="GridView1" Name="DummyID" 
    
                PropertyName="SelectedValue" Type="Object" />
    
        </WhereParameters>
    
    </asp:LinqDataSource>

In the snippet above, you can see that the GridView1.SelectedValue is defined as a WhereParameter for the LinqDataSource. DummyID is the primary key of the data in GridView1 and is of type Guid. In the ControlParameter, its Type is set to Object because Guid is not available in the TypeCode enum. Unfortunately, this results in the linq expression parser treating the value as type String causing the comparison to fail.

According to the Dynamic Expression API, a flavor of which is used by the LinqDataSource internally, we can perform explicit conversions using the syntax type(expr) where type is a type name optionally followed by ? and expr is an expression. The expression language defines the following primitive types:

Object Boolean Char String SByte Byte
Int16 UInt16 Int32 UInt32 Int64 UInt64
Decimal Single Double DateTime TimeSpan Guid

The primitive types correspond to the similarly named types in the System namespace of the .NET Framework Base Class Library. You can also use the nullable form of a value type by writing a ? after the type name (ex: Where="Foo = Int32?(@Foo)").

Therefore, we can rewrite our where clause like so which gets rid of the exception.

    
    <asp:LinqDataSource ID="LinqDataSource2" runat="server" 
    
        ContextTypeName="DataClassesDataContext" 
    
        Select="new (DummyID, FirstName, LastName)" TableName="DummyTables" 
    
        Where="DummyID == Guid(@DummyID)" onselecting="LinqDataSource2_Selecting">
    
        <WhereParameters>
    
            <asp:ControlParameter ControlID="GridView1" Name="DummyID" 
    
                PropertyName="SelectedValue" Type="Object" />
    
        </WhereParameters>
    
    </asp:LinqDataSource>

IMO, both these exceptions and a couple of others could probably be avoided if the LinqDataSource had a CancelSelectOnNullParameter and if the TypeCode enum had included the primitive types listed above and nullable value types.

Comments are always welcome!