DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma;
	h=message-id:date:from:user-agent:mime-version:to:subject
	:content-type:content-transfer-encoding;
	b=VJxwK3m5B67H6FgyN5Ej5FWO+rG1xkL2Hh/LCXMtD53O8I0ZFCa16bafJlDIRVuFnG
	5YFBRig2Fceqyji7f4hM3WttLBCJxxcQkthdXOHsAhGiStf4BQ6Hwtjll/3t8SbqrCQk
	5AyG6lBCvJhrmgBlFZ9iKEwRlwGAhX6VLogVc=
Message-ID: <4C7A5E1E.6050205@gmail.com>
Date: Sun, 29 Aug 2010 16:18:22 +0300
From: Volodymyr Kostyrko <c.kworr@gmail.com>
User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; uk-UA;
	rv:1.9.2.8) Gecko/20100807 Thunderbird/3.1.2
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: array can be slow when joining?
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit

I don't clearly understand why this happens, but when I try join some 
tables using arrays I end up with:

=# explain select count(*) from urls JOIN rules ON urls.tag && rules.tag;
                                       QUERY PLAN 

---------------------------------------------------------------------------------------
  Aggregate  (cost=1356.27..1356.28 rows=1 width=0)
    ->  Nested Loop  (cost=20.33..1354.96 rows=523 width=0)
          ->  Seq Scan on rules  (cost=0.00..1.01 rows=1 width=37)
          ->  Bitmap Heap Scan on urls  (cost=20.33..1347.42 rows=523 
width=29)
                Recheck Cond: (urls.tag && rules.tag)
                ->  Bitmap Index Scan on url_tag_g  (cost=0.00..20.20 
rows=523 width=0)
                      Index Cond: (urls.tag && rules.tag)

Here tag is text[] with list of tags. Whole select takes 142 ms. It 
drops down to 42 ms when I add some conditions that strip result table 
to zero length.

What am I missing? Is there any other ways to overlap those ones? Or 
should I find "any other way"?

-- 
Sphinx of black quartz judge my vow.