Re: Indexes not used for "min()"

Your name : Patrick
Your email address : Valsecchi

System Configuration
---------------------
Architecture (example: Intel Pentium) : Intel Pentium

Operating System (example: Linux 2.0.26 ELF) : Linux 2.4.20

PostgreSQL version (example: PostgreSQL-7.3.3): PostgreSQL-7.3.3

Compiler used (example: gcc 2.95.2) : gcc 3.2

Please enter a FULL description of your problem:
------------------------------------------------

When doing the following query, it's obvious that postgres should use
the index:

stats=# explain select min(time) from call;
QUERY PLAN
------------------------------------------------------------------------
Aggregate (cost=49779.82..49779.82 rows=1 width=8)
-> Seq Scan on stb_call (cost=0.00..44622.06 rows=2063106 width=8)
(2 rows)

stats=# \d call
Table "public.call"
Column | Type | Modifiers
---------+--------------------------+-----------
a | integer | not null
b | integer | not null
time | timestamp with time zone | not null
d | character varying(4) |
e | character varying(4) |
f | character varying(4) |
g | character varying(15) | not null
h | character varying(7) |
i | smallint | not null
Indexes: call_time btree ("time")

stats=# select count(*) from call;
count
---------
2063106
(1 row)

If you have any question or comment, please contact me directly, I'm not
suscribed to the list.

On Tue, Jul 29, 2003 at 09:48:37 -0600,
"Valsecchi, Patrick" <patrick(dot)valsecchi(at)nagrastar(dot)com> wrote:
>
> When doing the following query, it's obvious that postgres should use
> the index:
>
> stats=# explain select min(time) from call;

This has been discussed on the mailing lists numerous times. If you are
interested in details you can search the archives. If you just want a faster
equivalent query use:
select time from call order by time limit 1;