Retrieving Identity or Autonumber Values (ADO.NET)

.NET Framework 4

A primary key in a relational database is a column or combination of columns that always contain unique values. Knowing the primary key value allows you to locate the row that contains it. Relational database engines, such as SQL Server, Oracle, and Microsoft Access/Jet support the creation of automatically incrementing columns that can be designated as primary keys. These values are generated by the server as rows are added to a table. In SQL Server, you set the identity property of a column, in Oracle you create a Sequence, and in Microsoft Access you create an AutoNumber column.

A DataColumn can also be used to generate automatically incrementing values by setting the AutoIncrement property to true. However, you might end up with duplicate values in separate instances of a DataTable, if multiple client applications are independently generating automatically incrementing values. Having the server generate automatically incrementing values eliminates potential conflicts by allowing each user to retrieve the generated value for each inserted row.

During a call to the Update method of a DataAdapter, the database can send data back to your ADO.NET application as output parameters or as the first returned record of the result set of a SELECT statement executed in the same batch as the INSERT statement. ADO.NET can retrieve these values and update the corresponding columns in the DataRow being updated.

Some database engines, such as the Microsoft Access Jet database engine, do not support output parameters and cannot process multiple statements in a single batch. When working with the Jet database engine, you can retrieve the new AutoNumber value generated for an inserted row by executing a separate SELECT command in an event handler for the RowUpdated event of the DataAdapter.

Note
An alternative to using an auto incrementing value is to use the NewGuid method of a Guid object to generate a GUID, or globally unique identifier, on the client computer that can be copied to the server as each new row is inserted. The NewGuid method generates a 16-byte binary value that is created using an algorithm that provides a high probability that no value will be duplicated. In a SQL Server database, a GUID is stored in a uniqueidentifier column which SQL Server can automatically generate using the Transact-SQL NEWID() function. Using a GUID as a primary key can adversely affect performance. SQL Server 2005 introduces support for the NEWSEQUENTIALID() function, which generates a sequential GUID that is not guaranteed to be globally unique but that can be indexed more efficiently.

Note

An alternative to using an auto incrementing value is to use the NewGuid method of a Guid object to generate a GUID, or globally unique identifier, on the client computer that can be copied to the server as each new row is inserted. The NewGuid method generates a 16-byte binary value that is created using an algorithm that provides a high probability that no value will be duplicated. In a SQL Server database, a GUID is stored in a uniqueidentifier column which SQL Server can automatically generate using the Transact-SQL NEWID() function. Using a GUID as a primary key can adversely affect performance. SQL Server 2005 introduces support for the NEWSEQUENTIALID() function, which generates a sequential GUID that is not guaranteed to be globally unique but that can be indexed more efficiently.

Retrieving SQL Server Identity Column Values

When working with Microsoft SQL Server, you can create a stored procedure with an output parameter to return the identity value for an inserted row. The following table describes the three Transact-SQL functions in SQL Server that can be used to retrieve identity column values.

Function	Description
SCOPE_IDENTITY	Returns the last identity value within the current execution scope. SCOPE_IDENTITY is recommended for most scenarios.
@@IDENTITY	Contains the last identity value generated in any table in the current session. @@IDENTITY can be affected by triggers and may not return the identity value that you expect.
IDENT_CURRENT	Returns the last identity value generated for a specific table in any session and any scope.

The following stored procedure demonstrates how to insert a row into the Categories table and use an output parameter to return the new identity value generated by the Transact-SQL SCOPE_IDENTITY() function.

CREATE PROCEDURE dbo.InsertCategory
  @CategoryName nvarchar(15),
  @Identity int OUT
AS
INSERT INTO Categories (CategoryName) VALUES(@CategoryName)
SET @Identity = SCOPE_IDENTITY()

The stored procedure can then be specified as the source of the InsertCommand of a SqlDataAdapter object. The CommandType property of the InsertCommand must be set to StoredProcedure. The identity output is retrieved by creating a SqlParameter that has a ParameterDirection of Output. When the InsertCommand is processed, the auto-incremented identity value is returned and placed in the CategoryID column of the current row if you set the UpdatedRowSource property of the insert command to UpdateRowSource.OutputParameters or to UpdateRowSource.Both.

If your insert command executes a batch that includes both an INSERT statement and a SELECT statement that returns the new identity value, then you can retrieve the new value by setting the UpdatedRowSource property of the insert command to UpdateRowSource.FirstReturnedRecord.

C++

JScript

Private Sub RetrieveIdentity(ByVal connectionString As String)
    Using connection As SqlConnection = New SqlConnection( _
       connectionString)

        ' Create a SqlDataAdapter based on a SELECT query.
        Dim adapter As SqlDataAdapter = New SqlDataAdapter( _
           "SELECT CategoryID, CategoryName FROM dbo.Categories", _
           connection)

        ' Create the SqlCommand to execute the stored procedure. 
        adapter.InsertCommand = New SqlCommand("dbo.InsertCategory", _
           connection)
        adapter.InsertCommand.CommandType = CommandType.StoredProcedure

        ' Add the parameter for the CategoryName. Specifying the
        ' ParameterDirection for an input parameter is not required.
        adapter.InsertCommand.Parameters.Add( _
          "@CategoryName", SqlDbType.NVarChar, 15, "CategoryName")

        ' Add the SqlParameter to retrieve the new identity value.
        ' Specify the ParameterDirection as Output.
        Dim parameter As SqlParameter = _
           adapter.InsertCommand.Parameters.Add( _
          "@Identity", SqlDbType.Int, 0, "CategoryID")
        parameter.Direction = ParameterDirection.Output

        ' Create a DataTable and fill it.
        Dim categories As DataTable = New DataTable
        adapter.Fill(categories)

        ' Add a new row.
        Dim newRow As DataRow = categories.NewRow()
        newRow("CategoryName") = "New Category"
        categories.Rows.Add(newRow)

        ' Update the database.
        adapter.Update(categories)

        Console.WriteLine("List All Rows:")
        Dim row As DataRow
        For Each row In categories.Rows
            Console.WriteLine("{0}: {1}", row(0), row(1))
        Next
    End Using
End Sub

Merging New Identity Values

A common scenario is to call the GetChanges method of a DataTable to create a copy that contains only changed rows, and to use the new copy when calling the Update method of a DataAdapter. This is especially useful when you need to marshal the changed rows to a separate component that performs the update. Following the update, the copy can contain new identity values that must then be merged back into the original DataTable. The new identity values are likely to be different from the original values in the DataTable. To accomplish the merge, the original values of the AutoIncrement columns in the copy must be preserved, in order to be able to locate and update existing rows in the original DataTable, rather than appending new rows containing the new identity values. However, by default those original values are lost after a call to the Update method of a DataAdapter, because AcceptChanges is implicitly called for each updated DataRow.

There are two ways to preserve the original values of a DataColumn in a DataRow during a DataAdapter update:

The first method of preserving the original values is to set the AcceptChangesDuringUpdate property of the DataAdapter to false. This affects every DataRow in the DataTable being updated. For more information and a code example, see AcceptChangesDuringUpdate.
The second method is to write code in the RowUpdated event handler of the DataAdapter to set the Status to SkipCurrentRow. The DataRow is updated but the original value of each DataColumn is preserved. This method enables you to preserve the original values for some rows and not for others. For example, your code can preserve the original values for added rows and not for edited or deleted rows by first checking the StatementType and then setting Status to SkipCurrentRow only for rows with a StatementType of Insert.

When either of these methods is used to preserve original values in a DataRow during a DataAdapter update, ADO.NET performs a series of actions to set the current values of the DataRow to new values returned by output parameters or by the first returned row of a result set, while still preserving the original value in each DataColumn. First, the AcceptChanges method of the DataRow is called to preserve the current values as original values, and then the new values are assigned. Following these actions, DataRows that had their RowState property set to Added will have their RowState property set to Modified, which may be unexpected.

How the command results are applied to each DataRow being updated is determined by the UpdatedRowSource property of each DbCommand. This property is set to a value from the UpdateRowSource enumeration.

The following table describes how the UpdateRowSource enumeration values affect the RowState property of updated rows.

Member name	Description
Both	AcceptChanges is called and both output parameter values and/or the values in the first row of any returned result set are placed in the DataRow being updated. If there are no values to apply, the RowState will be Unchanged.
FirstReturnedRecord	If a row was returned, AcceptChanges is called and the row is mapped to the changed row in the DataTable, setting the RowState to Modified. If no row is returned, then AcceptChanges is not called and the RowState remains Added.
None	Any returned parameters or rows are ignored. There is no call to AcceptChanges and the RowState remains Added.
OutputParameters	AcceptChanges is called and any output parameters are mapped to the changed row in the DataTable, setting the RowState to Modified. If there are no output parameters, the RowState will be Unchanged.

Example

This example demonstrates extracting changed rows from a DataTable and using a SqlDataAdapter to update the data source and retrieve a new identity column value. The InsertCommand executes two Transact-SQL statements; the first one is the INSERT statement, and the second one is a SELECT statement that uses the SCOPE_IDENTITY function to retrieve the identity value.

INSERT INTO dbo.Shippers (CompanyName) 
VALUES (@CompanyName);
SELECT ShipperID, CompanyName FROM dbo.Shippers 
WHERE ShipperID = SCOPE_IDENTITY();

The UpdatedRowSource property of the insert command is set to UpdateRowSource.FirstReturnedRow and the MissingSchemaAction property of the DataAdapter is set to MissingSchemaAction.AddWithKey. The DataTable is filled and the code adds a new row to the DataTable. The changed rows are then extracted into a new DataTable, which is passed to the DataAdapter, which then updates the server.

C++

JScript

Private Sub MergeIdentityColumns(ByVal connectionString As String)

    Using connection As SqlConnection = New SqlConnection( _
       connectionString)

        ' Create the DataAdapter
        Dim adapter As SqlDataAdapter = New SqlDataAdapter( _
          "SELECT ShipperID, CompanyName FROM dbo.Shippers", connection)

        ' Add the InsertCommand to retrieve new identity value.
        adapter.InsertCommand = New SqlCommand( _
            "INSERT INTO dbo.Shippers (CompanyName) " & _
            "VALUES (@CompanyName); " & _
            "SELECT ShipperID, CompanyName FROM dbo.Shippers " & _
            "WHERE ShipperID = SCOPE_IDENTITY();", _
            connection)

        ' Add the parameter for the inserted value.
        adapter.InsertCommand.Parameters.Add( _
           New SqlParameter("@CompanyName", SqlDbType.NVarChar, 40, _
           "CompanyName"))
        adapter.InsertCommand.UpdatedRowSource = UpdateRowSource.Both

        ' MissingSchemaAction adds any missing schema to 
        ' the DataTable, including identity columns
        adapter.MissingSchemaAction = MissingSchemaAction.AddWithKey

        ' Fill the DataTable.
        Dim shipper As New DataTable
        adapter.Fill(shipper)

        ' Add a new shipper. 
        Dim newRow As DataRow = shipper.NewRow()
        newRow("CompanyName") = "New Shipper"
        shipper.Rows.Add(newRow)

        ' Add changed rows to a new DataTable. This
        ' DataTable will be used by the DataAdapter.
        Dim dataChanges As DataTable = shipper.GetChanges()

        ' Add the event handler. 
        AddHandler adapter.RowUpdated, New _
           SqlRowUpdatedEventHandler(AddressOf OnRowUpdated)

        ' Update the datasource with the modified records.
        adapter.Update(dataChanges)

        ' Merge the two DataTables.
        shipper.Merge(dataChanges)

        ' Commit the changes.
        shipper.AcceptChanges()

        Console.WriteLine("Rows after merge.")
        Dim row As DataRow
        For Each row In shipper.Rows
            Console.WriteLine("{0}: {1}", row(0), row(1))
        Next
    End Using
End Sub

The OnRowUpdated event handler checks the StatementType of the SqlRowUpdatedEventArgs to determine if the row is an insert. If it is, then the Status property is set to SkipCurrentRow. The row is updated, but the original values in the row are preserved. In the main body of the procedure, the Merge method is called to merge the new identity value into the original DataTable, and finally AcceptChanges is called.

C++

JScript

Private Sub OnRowUpdated( _
    ByVal sender As Object, ByVal e As SqlRowUpdatedEventArgs)
    ' If this is an insert, then skip this row.
    If e.StatementType = StatementType.Insert Then
        e.Status = UpdateStatus.SkipCurrentRow
    End If
End Sub

Retrieving Microsoft Access Autonumber Values

This section includes a sample that shows how to retrieve Autonumber values from a Jet 4.0 database. The Jet database engine does not support the execution of multiple statements in a batch or the use of output parameters, so it is not possible to use either of these techniques to return the new Autonumber value assigned to an inserted row. However, you can add code to the RowUpdated event handler that executes a separate SELECT @@IDENTITY statement to retrieve the new Autonumber value.

Example

Instead of adding schema information using MissingSchemaAction.AddWithKey, this example configures a DataTable with the correct schema prior to calling the OleDbDataAdapter to fill the DataTable. In this case, the CategoryID column is configured to decrement the value assigned each inserted row starting from zero, by setting AutoIncrement to true, AutoIncrementSeed to 0, and AutoIncrementStep to -1. The code then adds two new rows and uses GetChanges to add the changed rows to a new DataTable that is passed to the Update method.

C++

JScript

Shared connection As OleDbConnection = Nothing

Private Shared Sub MergeIdentityColumns(ByVal connection As OleDbConnection)
    Using connection

        ' Create a DataAdapter based on a SELECT query.
        Dim adapter As OleDbDataAdapter = New OleDbDataAdapter( _
          "SELECT CategoryID, CategoryName FROM Categories", _
          connection)

        ' Create the INSERT command for the new category.
        adapter.InsertCommand = New OleDbCommand( _
          "INSERT INTO Categories (CategoryName) Values(?)", connection)
        adapter.InsertCommand.CommandType = CommandType.Text

        ' Add the parameter for the CategoryName.
        adapter.InsertCommand.Parameters.Add( _
          "@CategoryName", OleDbType.VarWChar, 15, "CategoryName")
        adapter.InsertCommand.UpdatedRowSource = UpdateRowSource.Both

        ' Create a DataTable.
        Dim categories As DataTable = New DataTable

        ' Create the CategoryID column and set its auto
        ' incrementing properties to decrement from zero.
        Dim column As New DataColumn()
        column.DataType = System.Type.GetType("System.Int32")
        column.ColumnName = "CategoryID"
        column.AutoIncrement = True
        column.AutoIncrementSeed = 0
        column.AutoIncrementStep = -1
        categories.Columns.Add(column)

        ' Create the CategoryName column.
        column = New DataColumn()
        column.DataType = System.Type.GetType("System.String")
        column.ColumnName = "CategoryName"
        categories.Columns.Add(column)

        ' Set the primary key on CategoryID.
        Dim pKey(1) As DataColumn
        pKey(0) = categories.Columns("CategoryID")
        categories.PrimaryKey = pKey

        ' Fetch the data and fill the DataTable.
        adapter.Fill(categories)

        ' Add a new row.
        Dim newRow As DataRow = categories.NewRow()
        newRow("CategoryName") = "New Category"
        categories.Rows.Add(newRow)

        ' Add another new row.
        Dim newRow2 As DataRow = categories.NewRow()
        newRow2("CategoryName") = "Another New Category"
        categories.Rows.Add(newRow2)

        ' Add changed rows to a new DataTable that will be
        ' used to post the inserts to the database.
        Dim dataChanges As DataTable = categories.GetChanges()

        ' Include an event to fill in the Autonumber value.
        AddHandler adapter.RowUpdated, _
          New OleDbRowUpdatedEventHandler(AddressOf OnRowUpdated)

        ' Update the database, inserting the new rows.
        adapter.Update(dataChanges)

        Console.WriteLine("Rows before merge:")
        Dim row1 As DataRow
        For Each row1 In categories.Rows
            Console.WriteLine("  {0}: {1}", row1(0), row1(1))
        Next

        ' Merge the two DataTables.
        categories.Merge(dataChanges)

        ' Commit the changes.
        categories.AcceptChanges()

        Console.WriteLine("Rows after merge:")
        Dim row As DataRow
        For Each row In categories.Rows
            Console.WriteLine("  {0}: {1}", row(0), row(1))
        Next
    End Using
End Sub

The RowUpdated event handler uses the same open OleDbConnection as the Update statement of the OleDbDataAdapter. It checks the StatementType of the OleDbRowUpdatedEventArgs for inserted rows. For each inserted row a new OleDbCommand is created to execute the SELECT @@IDENTITY statement on the connection, returning the new Autonumber value, which is placed in the CategoryID column of the DataRow. The Status property is then set to UpdateStatus.SkipCurrentRow to suppress the hidden call to AcceptChanges. In the main body of the procedure, the Merge method is called to merge the two DataTable objects, and finally AcceptChanges is called.

C++

JScript

Private Shared Sub OnRowUpdated( _
    ByVal sender As Object, ByVal e As OleDbRowUpdatedEventArgs)
    ' Conditionally execute this code block on inserts only.
    If e.StatementType = StatementType.Insert Then
        ' Retrieve the Autonumber and store it in the CategoryID column.
        Dim cmdNewID As New OleDbCommand("SELECT @@IDENTITY", _
           connection)
        e.Row("CategoryID") = CInt(cmdNewID.ExecuteScalar)
        e.Status = UpdateStatus.SkipCurrentRow
    End If
End Sub