Message-ID: <3E3A9E49.1090503@gingerall.cz>
Date: Fri, 31 Jan 2003 17:03:21 +0100
From: Pavel Hlavnicka <pavel@gingerall.cz>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2) Gecko/20021203
MIME-Version: 1.0
To: pgsql-sql@postgresql.org
Cc: Tom Lane <tgl@sss.pgh.pa.us>
Subject: Re: calculated expressions and index use
References: <3E394AA6.8080400@gingerall.cz> <20100.1043945485@sss.pgh.pa.us>
	<3E3A3757.3040906@gingerall.cz> <8577.1044024965@sss.pgh.pa.us>
In-Reply-To: <8577.1044024965@sss.pgh.pa.us>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

Yes, you are right. It was an issue reated to the type compatibility. My 
braindead script created tables with 'datetime' types, what is 
'timestamp without time zone' after I switched to 'timestamp', all works 
fine.

Thanks a loc, the fact, you ensured me, that the constant expression 
should be recognized, pushed me much further.

Thanks again

Have a nice day

Pavel

Tom Lane wrote:
> Pavel Hlavnicka <pavel@gingerall.cz> writes:
> 
>>I use 7.2.1. :(
> 
> 
> 7.2 certainly will fold "'now'::timestamp - '1 hour'::interval" to a
> timestamp constant.  This could be a datatype compatibility issue
> (is created_at the same type?  in particular, with/without time zone?)
> or it could be a selectivity issue --- ie, the planner might think the
> condition covers too much of the table to make the index useful.
> 
> It would be useful to look at the results of EXPLAIN ANALYZE for this
> query, with and without ENABLE_SEQSCAN turned off.
> 
> 			regards, tom lane

-- 
Pavel Hlavnicka
Ginger Alliance
www.gingerall.com