How often do you see DataReader code that looks like this?
using (IDataReader dr = cmd.ExecuteNonQuery())
{
while (dr.Read())
{
int orderId = dr.GetInt32(0);
string customerId = dr.GetString(1);
int employeeId = dr.GetInt32(2);
DateTime orderDate = dr.GetDateTime(3);
double freight = dr.GetDouble(4);
// do stuff
}
dr.Close();
}If that looks like your code, you are not alone :-). Try searching codesearch.google.com for the following:
lang:c# (reader|dr)\.GetInt32\(
With the above code, anytime the order of columns in the SQL statement or stored procedure changes, the code is broken. And if you have lots of columns to read from, it’s a real nightmare to maintain the indexes.
The next time you write another DataReader loop, consider doing it this way instead:
using (IDataReader dr = cmd.ExecuteNonQuery())
{
int ORDER_ID = dr.GetOrdinal("OrderID");
int CUSTOMER_ID = dr.GetOrdinal("CustomerID");
int EMPLOYEE_ID = dr.GetOrdinal("EmployeeID");
int ORDER_DATE = dr.GetOrdinal("OrderDate");
int FREIGHT = dr.GetDecimal("Freight");
while (dr.Read())
{
int orderId = dr.GetInt32(ORDER_ID);
string customerId = dr.GetString(CUSTOMER_ID);
int employeeId = dr.GetInt32(EMPLOYEE_ID);
DateTime orderDate = dr.GetDateTime(ORDER_DATE);
double freight = dr.GetDouble(FREIGHT);
// do stuff
}
dr.Close();
}Using GetOrdinal makes the code much more readable and maintainable. You are calling GetOrdinal just once for each column, any performance penalty is insignificant compared to the benefits. Be careful not to put the GetOrdinal code inside the while block as that will unnecessarily slow you down (about 3% according to this article).
