Question

I have an SSRS grid report with a header row that SUM()s select child rows.

In the SQL query, one Shift table inner joins to many Parts tables, meaning an Hours field on the Shift table is duplicated per part. I use the “Hide duplicates” SSRS feature on the Hours cell to fix this visually. The issue is that although the duplicates are hidden, the SUM() function (on the report, not the query) still includes them in its summation of hours.

So, I might have the following table and header. Duplicates of Hours (and Shift) are hidden, but still (undesirably) included in the summation.

SHIFT   HOURS     PART   ON HAND
=====   =======   ====   =======
        SUM:  8          SUM: 90
=====   =======   ====   =======
1st           3   X12Z        20
                  J23Z        10
2nd           1   Y36P        30
3rd           1   Q90F        30

I am simplifying the actual report, so displaying the data in a different way isn’t an option. Also, these are already in a group, so I’m not sure I can add another group. How do I get the SUM() function to return the sum of the hours shown?

EDIT:

I have posted a thorough answer to my own question, but if there is a better way to do this or my answer could be improved, please feel free to offer your critique. If there are still answers that significantly improve on my method, I will accept them.

Answer 1

Given how presumably common this format on a report is, I expected there to be more answers or at least a better way to do this. But, alas, my method follows:

Rather than implement a cursor and loop through the table, run this update on the original results in a stored procedure. It clears the values of duplicated fields (i.e., shift, hours) where the current row’s part isn’t the part of the first row returned for this Shift. If there is an order by in the original query, include it in the nested select:

update @tempResults
set    shift = null, -- Shift is a duplicated field.
       hours = null, -- Hours is a duplicated field.
where  part <> -- The part number identifies the duplicates.
       (
           select top 1 part
           from  #tempresults t
           where #tempresults.shift = t.shift -- Parts were grouped by shift.
           -- order by    the same fields from the original select statement.
       )

Essentially, the formula is to:

1. Nullify the fields in the “parent” table (the 1 in the 1:m relationship).
2. Set the update‘s where using the “child” table’s identifying field.
3. Set the select‘s where using the parent table’s identifying field.