SQL Results Missing When Adding Columns - sql

Question

Trying to figure out an issue that is causing my SQL Server query to return no results. I have a query which calls out where there is more than one unique 'requested delivery date' at a size level on a single PO. I do this using COUNTand DISTINCT. It works perfect until I add the fields 'PO_ITEM_NUMBER' and 'REQ_DELIV_DATE'- this was requested by the business. I am not sure why that would cause an issue - for reference our tier for PO tables is Header,Item,Size - size being the most granular and it is SAP based.
Query:
SELECT E.TEAM_MEMBER_NAME [EMPLOYEE],
H.PO_TYPE,
H.PO_ISSUE_DATE,
S.PO_NUMBER,
S.MATERIAL,
M.DESCRIPTION,
H.PO_ORDERED_QUANTITY [PO_QUANTITY], -- if you sue SUM(S.PO_ORDERED_QUANTITY - you get more results but wrong totals
K.BUSINESS_SEGMENT_DESC,
S.PO_REQ_DELIV_DATE,
S.PO_ITEM_NUMBER
FROM PDX_SAP_USER..VW_PO_SIZE S --- you can use .. insetead of .dbo.
JOIN ADI_USER_MAINTAINED..SCM_PO_Employee_Name E --- join the po to employee assigment table
ON S.PO_NUMBER = E.PO_NUMBER
JOIN PDX_SAP_USER..VW_PO_HEADER H
ON E.PO_NUMBER = H.PO_NUMBER
JOIN PDX_SAP_USER..VW_PO_ITEM I
ON E.PO_NUMBER = I.PO_NUMBER
JOIN PDX_SAP_USER..VW_MM_MATERIAL M
ON E.MATERIAL = M.MATERIAL
JOIN PDX_SAP_USER..vw_kd_BUSINESS_SEGMENT K
ON M.BUSINESS_SEGMENT_CODE = K.BUSINESS_SEGMENT_CODE
WHERE I.PO_BALANCE_QUANTITY > 0 ---exclude any fully received PO's
AND NOT EXISTS (SELECT * FROM VW_PO_ITEM I1 WHERE DEL_INDICATOR = 'L' AND I.PO_NUMBER = I1.PO_NUMBER)
GROUP BY S.PO_NUMBER,
E.TEAM_MEMBER_NAME,
H.PO_TYPE,
H.PO_ISSUE_DATE,
S.MATERIAL,
M.DESCRIPTION,
K.BUSINESS_SEGMENT_DESC,
H.PO_ORDERED_QUANTITY,
S.PO_REQ_DELIV_DATE,
S.PO_ITEM_NUMBER
HAVING COUNT(DISTINCT S.PO_REQ_DELIV_DATE) > 1
ORDER BY S.PO_NUMBER
Adding query that works along with the results:
SELECT E.TEAM_MEMBER_NAME [EMPLOYEE],
H.PO_TYPE,
CONVERT(VARCHAR(12),H.PO_ISSUE_DATE,101) [PO_ISSUE_DATE],
S.PO_NUMBER,
S.MATERIAL,
M.DESCRIPTION,
H.PO_ORDERED_QUANTITY [PO_QUANTITY], --- if you use SUM(S.PO_ORDERED_QUANTITY) - you get more results but wrong totals
K.BUSINESS_SEGMENT_DESC
FROM PDX_SAP_USER..VW_PO_SIZE S --- you can use .. insetead of .dbo.
JOIN ADI_USER_MAINTAINED..SCM_PO_Employee_Name E --- join the po to employee assigment table
ON S.PO_NUMBER = E.PO_NUMBER
JOIN PDX_SAP_USER..VW_PO_HEADER H
ON E.PO_NUMBER = H.PO_NUMBER
JOIN PDX_SAP_USER..VW_PO_ITEM I
ON E.PO_NUMBER = I.PO_NUMBER
JOIN PDX_SAP_USER..VW_MM_MATERIAL M
ON E.MATERIAL = M.MATERIAL
JOIN PDX_SAP_USER..vw_kd_BUSINESS_SEGMENT K
ON M.BUSINESS_SEGMENT_CODE = K.BUSINESS_SEGMENT_CODE
WHERE I.PO_BALANCE_QUANTITY > 0 ---exclude any fully received PO's
AND NOT EXISTS (SELECT * FROM VW_PO_ITEM I1 WHERE DEL_INDICATOR = 'L' AND I.PO_NUMBER = I1.PO_NUMBER)
GROUP BY S.PO_NUMBER,
E.TEAM_MEMBER_NAME,
H.PO_TYPE,
H.PO_ISSUE_DATE,
S.MATERIAL,
M.DESCRIPTION,
K.BUSINESS_SEGMENT_DESC,
H.PO_ORDERED_QUANTITY
HAVING COUNT(DISTINCT S.PO_REQ_DELIV_DATE) > 1
ORDER BY S.PO_NUMBER
Results:
EMPLOYEE PO_TYPE PO_ISSUE_DATE PO_NUMBER MATERIAL DESCRIPTION PO_QUANTITY BUSINESS_SEGMENT_DESC
------------------------------ ------- ------------- ---------- ------------------ ---------------------------------------- --------------------------------------- ----------------------------------------------------------------------------------------------------
Christopher Olson NB 01/19/2017 0282238419 CD7078 ESS 3S PANT WVN 2054 CORE APP MEN SPORT ADIDAS
Juan Gomez NB 02/23/2017 0282524995 S98775 ESS LIN P/O FT 103 CORE APP MEN SPORT ADIDAS
Christopher Olson NB 03/09/2017 0282598957 BK7410 ESS LGO T P SJ 619 ATHLETICS APP MEN ADIDAS
Juan Gomez NB 03/28/2017 0282706115 S97155 ESS LIN TIGHT 961 CORE APP WOMEN SPORT ADIDAS
Juan Gomez NB 09/21/2017 0283752965 CF8152 BOS LABEL 7900 ATHLETICS APP MEN ADIDAS
Julie Lange-May 12 10/02/2017 0283796594 DQ1421 WOVEN JACKET W 1020 ATHLETICS APP WOMEN ADIDAS
Kekai Ariola NB 10/10/2017 0283837426 AC7366 PW HU HOLI Tennis Hu MC 5655 STATEMENT FTW ADIDAS
Cody Lofquist NB 11/10/2017 0283944933 DB2061 PREDATOR TANGO 18.1 TR 1756 FOOTBALL FTW ADIDAS
Andrew Zapata 05 11/13/2017 0283961402 CG6440 NEMEZIZ 18.1 FG W 543 FOOTBALL FTW ADIDAS
Christopher Olson NB 11/20/2017 0283981666 CV7748 ASSITA 17 GK Y 1648 FOOTBALL APP GENERIC ADIDAS
Cody Lofquist NB 11/21/2017 0283984539 DB2165 COPA 18.1 FG 501 FOOTBALL FTW ADIDAS
Julie Lange-May NB 11/26/2017 0284043157 CE4368 I GRPHC STSET 1333 ORIGINALS APP KIDS ADIDAS
Trey Pflug NB 11/27/2017 0284048754 CQ3168 SOLAR BOOST M 3500 RUNNING FTW MEN ADIDAS
Dave Laws NB 11/28/2017 0284059045 DB2966 YEEZY 500 15334 YEEZY FTW ADIDAS
Dave Laws NB 11/28/2017 0284059047 DB2966 YEEZY 500 12584 YEEZY FTW ADIDAS
Christopher Olson NB 12/06/2017 0284094060 BJ9165 TASTIGO17 SHO W 7522 FOOTBALL APP GENERIC ADIDAS
Christopher Olson NB 12/06/2017 0284094212 BK0350 TIRO17 TRG PNTW 7091 FOOTBALL APP GENERIC ADIDAS
Cody Lofquist NB 12/08/2017 0284107301 DB2062 PREDATOR TANGO 18.1 TR 2110 FOOTBALL FTW ADIDAS
Trey Pflug NB 12/11/2017 0284115640 BC0674 SOLAR BOOST W 1752 RUNNING FTW WOMEN ADIDAS
Kim Moreland NB 12/12/2017 0284137355 DJ3033 D2M K SHT 1730 CORE APP WOMEN SPORT ADIDAS
Cody Lofquist NB 12/12/2017 0284141196 DB2126 PREDATOR TANGO 18.3 IN 1988 FOOTBALL FTW ADIDAS
Cody Lofquist NB 12/12/2017 0284141253 AQ0612 NEMEZIZ MESSI TANGO 18.3 TF 526 FOOTBALL FTW ADIDAS
Dave Laws NB 12/15/2017 0284170426 DB2966 YEEZY 500 2918 YEEZY FTW ADIDAS
Cody Lofquist NB 12/16/2017 0284174671 DB2248 X 18.1 FG 668 FOOTBALL FTW ADIDAS
Cody Lofquist NB 12/16/2017 0284174673 DB2039 PREDATOR 18.1 FG 489 FOOTBALL FTW ADIDAS
Christopher Olson ER 12/20/2017 0284207872 BS4250 TASTIGO17 SHO 404 FOOTBALL APP GENERIC ADIDAS
Ben Paul NB 12/19/2017 0284208137 CG0584 REAL A JSY 811 FOOTBALL APP LICENSED ADIDAS
Julie Lange-May NB 01/07/2018 0284316616 DN4273 UAS BEANIE 120 ORIGINALS APP MEN ADIDAS
Cody Lofquist NB 01/08/2018 0284319552 DB2063 PREDATOR TANGO 18.1 TR 2001 FOOTBALL FTW ADIDAS
Cody Lofquist NB 01/19/2018 0284464341 DB2214 X 18+ FG 582 FOOTBALL FTW ADIDAS
Cody Lofquist NB 01/19/2018 0284464343 DB2013 PREDATOR 18+ FG 2201 FOOTBALL FTW ADIDAS
Cody Lofquist NB 01/19/2018 0284464344 DB2072 NEMEZIZ 18+ FG 1467 FOOTBALL FTW ADIDAS
Cody Lofquist NB 01/19/2018 0284464346 DB2251 X 18.1 FG 620 FOOTBALL FTW ADIDAS
Cody Lofquist NB 01/19/2018 0284464348 DB2167 COPA 18.1 FG 1714 FOOTBALL FTW ADIDAS
Cody Lofquist NB 01/19/2018 0284464349 DB2089 NEMEZIZ MESSI 18.1 FG 988 FOOTBALL FTW ADIDAS
Cody Lofquist NB 01/19/2018 0284464350 DB2040 PREDATOR 18.1 FG 2061 FOOTBALL FTW ADIDAS
Cody Lofquist NB 01/19/2018 0284465944 DB2001 PREDATOR 18.3 FG 7008 FOOTBALL FTW ADIDAS
Cody Lofquist NB 01/23/2018 0284489924 772109 SAMBA CLASSIC 419 FOOTBALL FTW ADIDAS
Andrew Zapata 05 02/02/2018 0284539184 DH3869 CLIMA 3.0 TEE 1853 ACTION SPORTS APP ADIDAS
Cody Lofquist NB 02/06/2018 0284550445 BB0571 Goletto VI FG J 5562 FOOTBALL FTW ADIDAS
Cody Lofquist NB 02/24/2018 0284666220 DM2092 MLS ASG OMB 424 FOOTBALL ACC HW ADIDAS
Christopher Olson NB 01/19/2018 0284666914 BP9111 D2M 3S SHORT 11811 CORE APP MEN SPORT ADIDAS
Cody Lofquist NB 02/27/2018 0284684097 019228 MUNDIAL TEAM 657 FOOTBALL FTW ADIDAS
Cody Lofquist NB 03/06/2018 0284704098 CD4683 GENERICWCBOX 80000 FOOTBALL ACC HW ADIDAS
Kekai Ariola NB 03/08/2018 0284728508 BB7619 Sobakov 1775 ORIGINALS FTW MEN ADIDAS
Kim Moreland NB 03/08/2018 0284730274 BP9733 ULT SS T 2557 TRAINING APP MEN ADIDAS
Kekai Ariola 05 03/27/2018 0284865999 B37532 EQT SUPPORT SK PK W 347 ORIGINALS FTW WOMEN ADIDAS
Kekai Ariola 05 03/27/2018 0284866000 B37545 EQT SUPPORT SK PK W 357 ORIGINALS FTW WOMEN ADIDAS
Kim Moreland NB 04/06/2018 0284914322 DH3591 Tech Tee 10042 TRAINING APP WOMEN ADIDAS
Cody Lofquist NB 04/10/2018 0284930265 CW5627 Pred FS JR MN 1651 FOOTBALL ACC HW ADIDAS
Kekai Ariola NB 04/10/2018 0284930449 B41794 PW TENNIS HU 315 ORIGINALS FTW MEN ADIDAS
(51 row(s) affected)

Answer 1

If you added HAVING clause recently then comment and check.
If not then change the INNER JOIN TO LEFT join from bottom to top one at a time and execute the query and check.
This way you can troubleshoot the issue easily.
Unless we have the sample data for all the tables you are used the query it will be difficult to answer.

Answer 2

WITH POsToReturn AS (
SELECT S.PO_NUMBER
FROM PDX_SAP_USER..VW_PO_SIZE S
JOIN PDX_SAP_USER..VW_PO_ITEM I
ON I.PO_NUMBER = S.PO_NUMBER
WHERE I.PO_BALANCE_QUANTITY > 0
GROUP BY S.PO_NUMBER
HAVING COUNT(DISTINCT S.PO_REQ_DELIV_DATE) > 1
AND COUNT(CASE WHEN I.DEL_INDICATOR = 'L' THEN 1 END) = 0
)
SELECT <your columns>
FROM POsToReturn P
JOIN PDX_SAP_USER..VW_PO_SIZE S
ON S.PO_NUMBER = P.PO_NUMBER
... <join the rest of the tables for the detail columns>
-- leave out the entire group by!
Or you could go with an approach like this:
with data as (
select
count(distinct S.PO_REQ_DELIV_DATE)
over (partition by S.PO_NUMBER) as rdd_count
<insert rest of main query>
)
select ... from data
where rdd_count > 1
order by S.PO_NUMBER;
As for understanding why, I think you've seen other queries where people added columns to a group by clause so they could get around a mysterious error message about non-aggregates. If not that then you've seen those systems that confusingly permit a non-standard behavior that gives results without erroring at all.
Usually what people want to accomplish is something like this: I'm grouping on Customer ID already but I want Customer Name in the results too. So they add that extra column to the 'group by' list and everything works fine. But if you think about it the reason is that the new column didn't change the groups at all since each Customer ID always has the same Customer Name anyway and it's ultimately just an easy way to get rid of the error. In your query that's not true though: you do have more than one date in all the groups you care about.
In my opinion it's better to use dummy aggregates like min(Customer Name) as CustomerName. Remember an aggregate function's purpose is collapsing multiple values into a single value. When necessary some systems will just pick a value at random, without warning you it did that. Many MySQL and Sybase developers got burned by this when they relied on this quirky behavior and/or never really learned how it's supposed to work.
Also remember that in general a group is a set of multiple rows collapsed into just one row. For count(distinct) to work it needed to see multiple dates across a single group. But that conflicts with your need to keep the output as separate rows.
Essentially the row by itself doesn't give you enough information to decide whether to keep it. I solved that by using a second query to figure a list of PO_Numbers meeting the criteria and then using that as a filter (via inner join.)
In the second example I used a window function instead, which lets you look outside each row without the row-collapsing behavior of group by. Both of the basically let you do what the having was intended to accomplish.