Sql Order by on multiple column

问题

I've the below result

VendorName | IncidentID | IncidentStatus | IncidentDate
-------------------------------------------------------
XYZ        | 100        |     Open       | 02-JUN-2011    
XYZ        | 101        |     Open       | 03-JUN-2011  
ABC        | 102        |     Open       | 01-JUN-2011  
XYZ        | 103        |     Open       | 01-APR-2011  
ABC        | 105        |     Open       | 05-JUN-2011

I want to order VendorName which has latest incident. Vendor ABC has the latest incident hence it should come first with all other incident for same vendor and then next Vendor with all respective incident in descending order.The desired result is like this -

VendorName | IncidentID | IncidentStatus | IncidentDate  
-------------------------------------------------------
ABC        | 105        |     Open       | 05-JUN-2011 
ABC        | 102        |     Open       | 01-JUN-2011
XYZ        | 101        |     Open       | 03-JUN-2011 
XYZ        | 100        |     Open       | 02-JUN-2011    
XYZ        | 103        |     Open       | 01-APR-2011

ORDER BY IncidentDate desc, VendorName doesn't give the desired output. Any help ?

回答1:

Use analytic functions:

SELECT *
FROM(
    SELECT 
        VendorName, 
        IncidentID, 
        IncidentStatus, 
        IncidentDate, 
        MAX(IncidentDate) OVER (PARTITION BY VendorName) maxDate
    FROM yourTable
) t
ORDER BY t.maxDate DESC, t.VendorName ASC, t.IncidentDate DESC

Refer to: http://docs.oracle.com/javadb/10.8.2.2/ref/rrefsqlj13658.html http://docs.oracle.com/cd/E11882_01/server.112/e10592/functions003.htm http://docs.oracle.com/cd/E11882_01/server.112/e26088/functions004.htm

回答2:

This will do it ...

ORDER BY MAX(INCIDENTDATE) OVER (PARTITION BY VENDORNAME) DESC, INCIDENTDATE DESC

... but I'm not sure if the analytic function is allowed in the ORDER BY. If it isn't, calculate it in a subquery and order by in the main query ...

select ...
from   (
  select Max(incidentdate) over (partition by vendorname) max_incidentdate_by_vendor,
         ...)
order by max_incidentdate_by_vender desc, incidentdate desc

回答3:

If you are on a RAC installation

set linesize 300
column REDOLOG_FILE_NAME format a50
SELECT
    a.INST_ID,
    a.GROUP#,
    a.THREAD#,
    a.SEQUENCE#,
    a.ARCHIVED,
    a.STATUS,
    b.MEMBER    AS REDOLOG_FILE_NAME,
    (a.BYTES/1024/1024/1024) AS SIZE_GB
FROM gv$log a
JOIN gv$logfile b ON a.Group#=b.Group# 
AND a.INST_ID=b.INST_ID 
ORDER BY a.INST_ID ASC, a.GROUP# ASC;

回答4:

select vendorname, incidentid, incidentstatus, incidentdate, max(incidentdate) 
over (partition by vendorname order by incidentdate desc) max_incidentdate
from t1 order by max_incidentdate desc, incidentdate desc

来源：https://stackoverflow.com/questions/13897024/sql-order-by-on-multiple-column

标签

sql

Oracle

sql-order-by