Question

I have the following data from 2 tables Notes (left) and scans (right) :

Imagine the picker and packers were all varying, like you can have JOHN, JANE etc.

I need a query that outputs like so :

On a given date range :

Name - Picked (units) - Packed (units)
MASI - 15 - 21
JOHN - 21 - 32

etc.

I can’t figure out how to even start this, any tips will be helpful thanks.

Answer 1

Without a “worker” take that lists each Picker/Packer individually, I think you’d need something like this…

SELECT
      CASE WHEN action.name = 'Picker' THEN scans.Picker ELSE scans.Packer END  AS worker,
  SUM(CASE WHEN action.name = 'Picker' THEN notes.Units  ELSE 0            END) AS PickedUnits,
  SUM(CASE WHEN action.name = 'Packer' THEN notes.Units  ELSE 0            END) AS PackedUnits
FROM
  notes
INNER JOIN
  scans
    ON scans.PickNote = notes.Number
CROSS JOIN
(
            SELECT 'Picker' AS name
  UNION ALL SELECT 'Packer' AS name

)
  AS action
GROUP BY
  CASE WHEN action.name = 'Picker' THEN scans.Picker ELSE scans.Packer END

(This is actually just an algebraic re-arrangement of the answer that @RaphaëlAlthaus posted at the same time as me. Both use UNION to work out the Picker values and the Packer values separately. If you have separate indexes on scans.Picker and scans.Packer then I would expect mine MAY be slowest. If you don’t have those two indexes then I would expect mine to be fastest. I recommend creating the indexes and testing on a realtisic data set.)

EDIT

Actually, what I would recommend is a change to scans table completely; normalise it.

• Your de-normalised set has one row per PickNote, with fields picker and packer.
• A normalised set would have two rows per PickNote with fields role and worker.

  id  | PickNote | Role | Worker
------+----------+------+--------
  01  | PK162675 | Pick |  MASI
  02  | PK162675 | Pack |  MASI
  03  | PK162676 | Pick |  FRED
  04  | PK162676 | Pack |  JOHN

This allows you to create simple indexes and simple queries.

You may initially baulk at the extra unecessary rows, but it will yield simpler queries, faster queries, better maintainability, increased flexibility, etc, etc.

In short, this normalisation may cost a little extra space, but it pays back dividends forever.