Question

All, I have the following table called Foo which is the output of a complex PIVOT query

I need to remove rows in this table which have (DB1, DB2, DB3) all NULL. To do this normally we could use something like

DELETE FROM Foo 
WHERE [DB1] IS NULL 
  AND [DB2] IS NULL 
  AND [DB3] IS NULL;

but the problem is that this code is called from C# and the columns (DB1, DB2, DB3) will change from execution-to-execution.

How can I do this dynamically?

Thanks for your time.

Edit1. I have another table Bar that contains the avalible DB# columns, so it is possible to use this table SELECT DISTINCT ... to get the list of columns I need in the WHERE clause. However, I have never done this before and am not sure where to start… Any help is appreciated.

Edit2. I cannot use C# in any way. The C# code is a parser of SQL, and works with the given .sql file. This has to be done entirely in SQL.

Answer 1

sp_executesql allows you to execute a SQL string built up in SQL script.

http://msdn.microsoft.com/en-us/library/ms188001.aspx

From your bar table

create proc DumpFooJunk 
as
begin 
    declare @strCols nvarchar(max)
    select @strCols = N''
    select @strCols = @strCols + ' AND ' + name + ' IS NULL ' from [bar]
    select @strCols = N'DELETE FROM [foo] WHERE ' + substring(@strCols,5,LEN(@strcols))
    select @strCols
    -- exec sp_executesql @strCols
end