Problems with filter on timestamp

From: Alex Turner <armtuk(at)gmail(dot)com>
To: PgSQL General List <pgsql-general(at)postgresql(dot)org>
Subject: Problems with filter on timestamp
Date: 2005-02-02 18:52:56
Message-ID: 33c6269f05020210527ace7710@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

I have a database that logs website requests, and I'm trying to get
all unique visitors within the last 1 minute, 5 minutes and 10
minutes. I have a table that I have the schema for below that incude
a field remote_ip.

When I perform the query with the filter, and the query without, the
results are the same:

dblogger=# select count(distinct(remote_ip)) from weblog_entry where
request_time>(current_time-interval '1 minute');
count
-------
313
(1 row)

dblogger=# select count(distinct(remote_ip)) from weblog_entry;
count
-------
316
(1 row)

dblogger=# explain select count(distinct(remote_ip)) from weblog_entry
where request_time>(current_time-interval '1 minute');
QUERY PLAN
-----------------------------------------------------------------------------------------------------------------
Aggregate (cost=951.98..951.98 rows=1 width=40)
-> Seq Scan on weblog_entry (cost=0.00..940.85 rows=4452 width=40)
Filter: ((request_time)::text > ((('now'::text)::time(6) with
time zone - '00:01:00'::interval))::text)
(3 rows)

This explain plain seems to me to be saying that it's casting the
timestamp to a text type before comparing with my
current_time-interval.

If I cast request_time directly to a time(6), then it works:

dblogger=# explain select count(distinct(remote_ip)) from weblog_entry
where request_time::time(6)>(current_time-interval '1 minute');

QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------------------------------------
Aggregate (cost=1177.00..1177.00 rows=1 width=40)
-> Seq Scan on weblog_entry (cost=0.00..1165.46 rows=4612 width=40)
Filter: (((request_time)::time(6) without time zone)::time
with time zone > (('now'::text)::time(6) with time zone -
'00:01:00'::interval))
(3 rows)

dblogger=# select count(distinct(remote_ip)) from weblog_entry where
request_time::time(6)>(current_time-interval '1 minute');
count
-------
18
(1 row)

Is this the right behaviour?

I'm using Postgresql 8.0.1 on AMD64 compiled from source with -m64
-march=k8 -O3 on SuSe 9.2

Alex Turner
NetEconomist

Responses

Browse pgsql-general by date

  From Date Subject
Next Message elein 2005-02-02 19:06:41 Re: plpython.so
Previous Message Tom Lane 2005-02-02 18:49:24 Re: plpython.so