Question

This may be a pipe dream, but I am trying to get these 2 commands that work separately, to work as one single command. Instead of XXXXX, I would like to dump in the select statement (which gives me the difference of yesterday’s total and today’s total). The combined command would find the most recent row (max idtanklevel) and update it with the difference of itself and the prior row.

update fw_db.tanklevel2 T3
inner join (select max(idtanklevel) as idtanklevel from fw_db.tanklevel2) T4
on T3.idtanklevel = T4.idtanklevel
set glycolsmallchange = XXXXX;

and

select round(sum((T1.glycolsmall) - (T2.glycolsmall)),2) 
from fw_db.tanklevel2 T1, fw_db.tanklevel2 T2 
where 
T1.idtanklevel = (select max(idtanklevel) from fw_db.tanklevel2)
and 
T2.idtanklevel=(select max(idtanklevel)-1 from fw_db.tanklevel2);

Thanks in advance.

Answer 1

Here’s how I would get all that done in a single statement. (Note that this query assumes that idtanklevel is the PRIMARY KEY or at least a UNIQUE KEY in the tanklevel2 table; the query will need to be adjusted if that’s not the case.)

Some brief comments to document the purpose of each table reference and each inline view:

-- m  = get id of latest row
-- t  = get latest row (the row to be updated)
-- pr = get id for the prior row
--      pri = source for prior row id
--      l = id of latest row (identical to "m")
-- pv = get value from prior row
-- p  = wrapper so we can join to row returned from pv

UPDATE fw_db.tanklevel2 t
  JOIN ( SELECT MAX(idtanklevel) AS idtanklevel FROM fw_db.tanklevel2 ) m
    ON m.idtanklevel = t.idtanklevel
 CROSS
  JOIN ( SELECT pv.glycolsmall
              , pv.idtanklevel
           FROM ( SELECT MAX(pri.idtanklevel) AS idtanklevel
                    FROM fw_db.tanklevel2 pri
                   WHERE pri.idtanklevel <
                         (SELECT MAX(l.idtanklevel) FROM fw_db.tanklevel2 l)
                ) pr
           JOIN fw_db.tanklevel2 pv
             ON pv.idtanklevel = pr.idtanklevel
          LIMIT 1
       ) p
   SET t.glycolsmallchange = ROUND(t.glycolsmall - p.glycolsmall,2)

Note that the query above uses a slightly different technique to identify the “prior” row. Rather than subtracting 1 from the maximum id value, the query gets the second highest id value (the highest id value that is lower than the maximum value).

To implement the same strategy your queries are using (i.e. subtracting 1 from the maximum id value, remove this portion of that query above:

           FROM ( SELECT MAX(pri.idtanklevel) AS idtanklevel
                    FROM fw_db.tanklevel2 pri
                   WHERE pri.idtanklevel <
                         (SELECT MAX(l.idtanklevel) FROM fw_db.tanklevel2 l)
                ) pr

and replace it with this:

           FROM ( SELECT MAX(l.idtanklevel)-1 AS idtanklevel
                    FROM fw_db.tanklevel2 l
                ) pr

---

To test this statement, without actually performing an update, remove the SET clause and replace the UPDATE keyword with SELECT and a list of relevant expressions that let you see the rows that are returned e.g.:

SELECT t.glycolsmallchange                    AS old_glycolsmallchange
     , ROUND(t.glycolsmall - p.glycolsmall,2) AS new_glycolsmallchange
     , t.idtanklevel  AS latest_row_id
     , t.glycolsmall  AS latest_glycolsmall
     , p.idtanklevel  AS prior_row_id
     , p.glycolsmall  AS prior_glycolsmall
  FROM