Whoops, a litle fast on the keyboard.
This bit about:
>I mean, isn't this really equivalent to:
>
>explain select max(event_id) from (select event_id from syslog_event order
>by event_id desc limit 1) as syslog_event;
>NOTICE: QUERY PLAN:
>
>Aggregate (cost=0.70..0.70 rows=1 width=8)
> -> Subquery Scan syslog_event (cost=0.00..0.70 rows=1 width=8)
> -> Limit (cost=0.00..0.70 rows=1 width=8)
> -> Index Scan Backward using syslog_event_pkey on syslog_event (cost=0.00..2435803.36 rows=3490163
width=8)
>
can obviously be shortened to:
explain select event_id from syslog_event order by event_id desc limit 1;
NOTICE: QUERY PLAN:
Limit (cost=0.00..0.70 rows=1 width=8)
-> Index Scan Backward using syslog_event_pkey on syslog_event (cost=0.00..2435803.36 rows=3490163 width=8)
Ed
which flies as expected? Now, this type of thing gets me real worried
about how good the optimizer really is. I have a number of fairly
complicated queries that are created via criteria from the web, and it
would be a pain to get them all "hand-optimized" if I cannot rely on the
optimizer at least picking reasonable methods.
Miscellaneous: Using 7.2.1 under OpenBSD 3.0
Ed