SqlBulkCopy class in C# is changing a numeric value

Question

I am using SqlBulkCopy.WriteToServer -method to insert bulk data to database table. When I am passing a datatable as the parameter of the method WriteToServer , some numeric values (numeric(17,6)) is getting subtracted by .000001 while it is inserted in database table. But not all values are getting changed, it is happening randomly, couldn't identify the pattern. eg- 10068.121 is being changed to 10068.120999

Now when I am passing a datareader as the parameter of the method WriteToServer, it is working fine, the numeric values remain unchanged.

Any idea why it is happening for datatable and not for datareader? My application is a C# console application.

Here is my code… if I pass the datatable (_dt) to _sqlBulkCopy.WriteToServer method, then I have the issue. But when I convert the datatable (_dt) to DataTableReader (_Reader), it is working fine.

DataTable _dt = new DataTable();
_odbCDA.Fill(_dt);
DataTableReader _Reader = new DataTableReader(_dt);

using (SqlBulkCopy _sqlBulkCopy = new SqlBulkCopy(_sqlConnectionString)) 
{
     _sqlBulkCopy.BulkCopyTimeout = 3600;
     _sqlBulkCopy.DestinationTableName = _tableName;
     _sqlBulkCopy.WriteToServer(_Reader); //This works
     //_sqlBulkCopy.WriteToServer(dt); //This is having the issue
     _sqlBulkCopy.Close();
}

Answer 1

It's hard to be certain without an example reproduction, but if I had to guess when you're using DataReader you're copying data between two tables that have the same definition/precision for the numeric type, but when using DataTable the column does not have matching precision. Presumably this is because the DataReader isn't mapping things into .NET types, but the DataTable is.

I think (but have not tested) .NET types map about like so (at least in SQL Server):

• C# float -> SQL real
• C# double -> SQL float
• C# decimal -> SQL money or numeric

SQL let's you specify precision in ways C# doesn't, so if you're working with C# built-in types there's a solid chance you're performing conversions to and from numeric(17,6) which could lose precision.

If my guess is correct, the DataType on the affected DataColumn in the DataTable will be something like System.Double , System.Single , or System.Decimal .