Generating SubTotals using GROUPING

  • Hugo Kornelis (7/20/2010)

    My final comment in this topic, and then I'll respect R.P.'s request.

    Jeff Moden (7/19/2010)

    Still, the order of run return is, in fact, documented in Books Online...


    Specifies that in addition to the usual rows provided by GROUP BY, summary rows are introduced into the result set. [font="Arial Black"]Groups are summarized in a hierarchical order[/font], from the lowest level in the group to the highest. [font="Arial Black"]The group hierarchy is determined by the order in which the grouping columns are specified. [/font]Changing the order of the grouping columns can affect the number of rows produced in the result set.

    This quote describes which possible summary rows are and are not introduced in the result set, not the order in which results are returned.

    (I'm also tempted to point out the use of the term "result set" rather than "recordset" in this quote, but the terminology in BOL as a whole is so often wrong that I'd probably better not go there)

    Jeff, or anyone else - if you want to continue this debate, then please start a new topic and send me a PM with the link, or include the link in this topic. Without a link, I'll probably never find the new topic.

    It's important that people on this thread realize that WITH ROLLUP and WITH CUBE produce the correct sorted results as stated in BOL. How much plainer can you get than [font="Arial Black"]Groups are summarized in a hierarchical order, from the lowest level in the group to the highest. [/font]

    --Jeff Moden

    RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
    First step towards the paradigm shift of writing Set Based code:
    ________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

    Change is inevitable... Change for the better is not.

    Helpful Links:
    How to post code problems
    How to Post Performance Problems
    Create a Tally Function (fnTally)

  • There are two things that I prefer to do in a different way:

    The first is your use of ISNULL(FieldName, 'ZTotal') -- I prefer to use a case statement

    CASE GROUPING(FieldName) WHEN 0 THEN FieldName ELSE 'Total' END -- this protects against a NULL value being mistaken for a total

    The second is your order by using 'ZTotal' to make it go to the bottom -- I prefer to once again use the GROUPING() function

    ORDER BY GROUPING(FieldName) -- which puts all the non-subtotals at the top since GROUPING() = 0 and the subtotals at the bottom where GROUPING() = 1

    This also allows you to easily reverse the order and put the totals at the top by adding DESC -- and you don't have to worry about the dreaded 'Zygote' which sorts lower than 'ZTotal'

    Another suggestion that I saw was using a Cube to get some the rollups to work in both directions -- you can always use a HAVING at the bottom to filter out any groups that you don't want, but you have to be very careful with the GROUP BY order or you'll lose more groupings


    HAVING GROUPING(FieldName) = 0 will not show any sub totals for that grouping

    HAVING GROUPING(FieldWithID) = GROUPING(FieldWithValue) will not show any extra grouping for fields like StateID and StateValue

  • For a prettier output, prefix the product and scheme results with a space (' '+....)

    Then you don't need to prefix the totals with a 'Z' as they will fall to the end anyway.

  • good article! and other comments from Hugo and gang. Just a comment, your query #2 is missing the column 'comment' on it.

  • Great article! My brain is finally starting to grasp what ROLLUP does.

    Rob Schripsema
    Propack, Inc.

  • Nice article.

Viewing 6 posts - 31 through 35 (of 35 total)

You must be logged in to reply to this topic. Login to reply