There is no guarantee that CHECKSUM_AGG() will return a unique value for every different recordset, so 'fixing' the proc is not really an option unless you are prepared to use...

...

Right now, my priority is to wait a few more days for the water to warm up. Predicting some time to relax and pull in some crappies. And maybe...

Could be the 'new' cardinality estimator (introduced in 2014). Lots of links all over, but you could start with this.

Please describe the problems you believe TRUNCATE introduces which...

I think this is right ...

As you have defined the datatype of the input parameter to be date, the error you see happens before the proc executes. You need to...

I understand your requirement, but unless you post the SQL you have tried, it's difficult to help you any further.

As far as I know, you have to delete and recreate precedence constraints. It's a pain, I agree.

Regarding your second post, is that a question?

Wouldn't this be unethical?

Do a left join from a table, or virtual, dynamic table, which contains all of the years you wish to display.

Hello Phil Thanks for your answers. - We have 50 tables that could be refreshed. Depending of the day, we receive data for a specific destination table. So column...

I agree that this is a horrid idea.

It is, however, pretty easy to solve. Put a Script task before each dataflow and add a wait there, possibly something like this...

Add a task to save the number of columns in the 'current' file before its data flow is executed.

Add a suitable precedence constraint to avoid running the data flow task...

Run the 'guts' of your stored proc, having modified the final SELECT such that it does a

SELECT INTO [Newtable] .....

FROM etc etc

WHERE 1 = 0  --Don't...

Do the column names change from day to day?

If they do, do you want to create a new table every day?

Or is the target table static in structure such that...

Thanks Thom, quite right! I should have added the text "in this case".