Håkan Jacobsson wrote:
Thanx Merlin, have a nice one (vacation)!
It turns out I have'nt described the problem accurately=(
Data may actually differ in two of the columns (the varchar
columns).
I still want to remove rows which share the same data in those
two columns and have the date column
set to NULL.
I.e. row 1,2,3 have:
column1 = 'foo', column2 = 'hey' and the date column =
NULL
row 4,5,6 have:
column1 = 'brat', column2 = 'yo' and the date column =
NULL
I want to keep just one of the 1 - 3 rows and one of the 4 - 6
rows..
I will try Merlins and Scotts solutions tomorrow. Anyone know
if I need to modify Merlins and/or Scotts
solutions to solve this new situation?
If i understand correctly, this should give you the records you want to
keep:
SELECT DISTINCT ON (t.one, t.two) t.one, t.two, t.three, [t.n] FROM foo
AS t;
Put those into a tmp table, truncate the original, then put the saved
rows back in.
brian
---------------------------(end of broadcast)---------------------------
TIP 4: Have you searched our list archives?
http://archives.postgresql.org/
