If you have the following function (for example):
CREATE FUNCTION dbo.UFN_SAMPLE_FUNCTION
(
@Param1 nvarchar(10),
@Param2 int = NULL
)
RETURNS TABLE
AS
RETURN
SELECT @Param1 AS Col1, @Param2 AS Col2;
GO
Then you can use it the following way (option 1):
SELECT * FROM dbo.UFN_SAMPLE_FUNCTION ('ABC', DEFAULT);
which is correct way and you get the following result:
Col1 Col2
---------- -----------
ABC NULL
But if you try to use parametrized query (option 2):
exec sp_executesql N'SELECT * FROM dbo.UFN_SAMPLE_FUNCTION (@P1, @P2)',N'@P1 nvarchar(10),@P2 int',@P1=N'abc',@P2=default;
you will get an error:
Msg 8178, Level 16, State 1, Line 0
The parameterized query '(@P1 nvarchar(10),@P2 int)SELECT * FROM dbo.UFN_SAMPLE_FUNCTION' expects the parameter '@P2', which was not supplied.
If you have the following .net code:
public void RunTVF(string param1, int? param2)
{
using (SqlConnection con = GetProdConection())
{
using (var cmd = new SqlCommand("SELECT * FROM dbo.UFN_SAMPLE_FUNCTION (@P1, @P2)", con))
{
cmd.CommandType = CommandType.Text;
var param = new SqlParameter
{
ParameterName = "@P1",
SqlDbType = SqlDbType.NVarChar,
Size = 10 ,
Value = param1
};
cmd.Parameters.Add(param);
param = new SqlParameter
{
ParameterName = "@P2",
SqlDbType = SqlDbType.Int,
Value = param2
};
cmd.Parameters.Add(param);
cmd.Connection.Open();
using (IDataReader dataReader = cmd.ExecuteReader())
{
//...
}
}
}
}
then, in case param2 = null as Jack suggested above, the script produced by the code will be identical to the option 2 and will result to the same error. So you cannot use NULL in this case.You cannot set DEFAULT as the value of SQLParameter either.
What you can do is to create a stored procedure to wrap the call to your funcion and move your default value from the function to the SP. Example:
CREATE PROCEDURE dbo.USP_SAMPLE_PROCEDURE
(
@Param1 nvarchar(10),
@Param2 int = NULL, --DEFAULT value now is here (remove it from the function)
@Statement nvarchar(max)
)
AS
BEGIN
SET NOCOUNT ON;
EXEC sp_executesql @Statement,N'@P1 nvarchar(10),@P2 int',@P1=@Param1,@P2=@Param2;
END
The .NET code will look the following way:
public void RunWrapper(string param1, int? param2)
{
using (SqlConnection con = GetProdConection())
{
using (var cmd = new SqlCommand("USP_SAMPLE_PROCEDURE", con))
{
cmd.CommandType = CommandType.StoredProcedure;
var param = new SqlParameter
{
ParameterName = "@Param1",
SqlDbType = SqlDbType.NVarChar,
Size = 10,
Value = param1
};
cmd.Parameters.Add(param);
param = new SqlParameter
{
ParameterName = "@Param2",
SqlDbType = SqlDbType.Int,
Value = param2
};
cmd.Parameters.Add(param);
param = new SqlParameter
{
ParameterName = "@Statement",
SqlDbType = SqlDbType.NVarChar,
Size = -1, //-1 used in case you need to specify nvarchar(MAX)
Value = "SELECT * FROM dbo.UFN_SAMPLE_FUNCTION (@P1, @P2)"
};
cmd.Parameters.Add(param);
cmd.Connection.Open();
using (IDataReader dataReader = cmd.ExecuteReader())
{
//...
}
}
}
}
In this case null as a value for the param2 will be translated to the correct DEFAULT and the following script will be produced:
exec USP_SAMPLE_PROCEDURE @Param1=N'ABC',@Param2=default,@Statement=N'SELECT * FROM dbo.UFN_SAMPLE_FUNCTION (@P1, @P2)'
which will give you the following result:
Col1 Col2
---------- -----------
ABC NULL
I am not sure that this is the best practice. This is just the work-around.
