Jeff Davis wrote: > On Mon, 2006-11-27 at 12:44 -0800, Ron Mayer wrote: >> Shouldn't the results of this query shown here been sorted by "b" rather than by "a"? >> >> I would have thought since "order by b" is in the outer sql statement it would have >> been the one the final result gets ordered by. >> >> li=# select * from (select (random()*10)::int as a, (random()*10)::int as b from generate_series(1,10) order by a) as x order by b; >> a | b >> ---+---- >> 0 | 8 >> 1 | 10 >> 3 | 4 >> 4 | 8 >> 5 | 1 >> 5 | 9 >> 6 | 4 >> 6 | 5 >> 8 | 4 >> 9 | 0 >> (10 rows) >>... > > It looks like a planner bug. > > Below are two plans; the first fails and the second succeeds. That leads > me to believe it's a planner bug, but what seems strangest to me is that > it does order by a, and not by some new evaluation of (random()*10). > Yeah, looks that way to me too. So how would I report it. Ccing the bugs list? Guess it can't hurt.
