I have "inherited" a little C# method that creates an ADO.NET SqlCommand object and loops over a list of items to be saved to the database (SQL Server 2005).
Right now, the traditional SqlConnection/SqlCommand approach is used, and to make sure everything works, the two steps (delete old entries, then insert new ones) are wrapped into an ADO.NET SqlTransaction.
using (SqlConnection _con = new SqlConnection(_connectionString))
{
using (SqlTransaction _tran = _con.BeginTransaction())
{
try
{
SqlCommand _deleteOld = new SqlCommand(......., _con);
_deleteOld.Transaction = _tran;
_deleteOld.Parameters.AddWithValue("@ID", 5);
_con.Open();
_deleteOld.ExecuteNonQuery();
SqlCommand _insertCmd = new SqlCommand(......, _con);
_insertCmd.Transaction = _tran;
// add parameters to _insertCmd
foreach (Item item in listOfItem)
{
_insertCmd.ExecuteNonQuery();
}
_tran.Commit();
_con.Close();
}
catch (Exception ex)
{
// log exception
_tran.Rollback();
throw;
}
}
}
Now, I've been reading a lot about the .NET TransactionScope class lately, and I was wondering, what's the preferred approach here? Would I gain anything (readibility, speed, reliability) by switching to using
using (TransactionScope _scope = new TransactionScope())
{
using (SqlConnection _con = new SqlConnection(_connectionString))
{
....
}
_scope.Complete();
}
What you would prefer, and why?
Marc
You won't immediately gain anything by switching your existing code to use TransactionScope
. You should use it for future development because of the flexibility it provides. It will make it easier in the future to include things other than ADO.NET calls into a transaction.
BTW, in your posted example, the SqlCommand
instances should be in using
blocks.