Calling and manipulating in PL/SQL

Question

Hi, got a problem. Lets say I had a table (TABLE T) with 2 columns (COL A) and (COL B) holding numerical data and I want to manipulate the data as follows:
COLA     COLB
  3          2
  4          4               
  5          7
  6          8
  4          5I want to double the values where the value directly above it is larger.
In the example, the last values of each column (4 and 5) are smaller than their predecessors (6 and 8) and so I want to change them to:
COLA     COLB
  3          2
  4          4               
  5          7
  6          8
  8          10In this case the two values that needed changing were in the same row but this is just coincidence.
I am stuck as to how I would I go about doing this in PL. I’m guessing it will entail an update of some sort with some conditions and a loop but I don’t know how to call a value from the row above and so am stuck as to how to start. I've been goofing around this problem so any suggestions would be much appreciated.

Answer 1

If you have to do this in "PL" (assuming you mean PL/SQL), an easy way would be to use a collection. Comparing individual elements is much easier with an index. But, that will be more like Greek to you, so the brute force approach is use a cursor for loop and store off the prior values into variables.
SQL> select * from t;
      COLA       COLB
         3          2
         4          4
         5          7
         6          8
         4          5
SQL> declare
  2    v_lastA number := -99;
  3    v_lastB number := -99;
  4    v_rowid urowid;
  5    cursor c is
  6    select rowid upd_rowid, colA, colB
  7    from t;
  8  begin
  9    for r in c loop
10      if v_lastA = 99 then
11        --we know this is the first record, so save
12        v_lastA := r.colA;
13        v_lastB := r.colB;
14      elsif
15        r.colA < v_lastA and r.colB < v_lastB then
16        --new values are strictly smaller, so do the update
17        update t
18        set colA = 2*r.colA,
19            colB = 2*r.colB
20        where rowid  = r.upd_rowid;
21      end if;
22      v_lastA := r.colA;
23      v_lastB := r.colB;
24    end loop;
25    commit;
26  end;
27  /
PL/SQL procedure successfully completed.
SQL>
SQL> select * from t;
      COLA       COLB
         3          2
         4          4
         5          7
         6          8
         8         10