To: pgsql-sql@postgresql.org
Path: not-for-mail
From: Jasen Betts <jasen@xnet.co.nz>
Newsgroups: gmane.comp.db.postgresql.sql
Subject: Re: UPDATE in a specific order
Date: 16 Dec 2010 19:58:50 GMT
Organization: Dis (not Dat) Organisation
Lines: 30
Message-ID: <iedr1q$dv1$1@reversiblemaps.ath.cx>
References: <4D0A28F0.3020205@planit.com.br>
NNTP-Posting-Host: localhost
User-Agent: slrn/pre1.0.0-18 (Linux)


> I need to make update of table1 with data on table2 in the order of id
> of table2

that looks like EAV. is it?

> I=B4m trying to do an update like this:

that's not going to work.

perhaps you can rewrite the from part to only return one row for every
table1_fk,  this one row will combine several rows from table2

> The "FROM ( SELECT * FROM table2 ORDER BY id  ) tHist"  is a try to
> force a specific order on table2 to update table1
> but this isn=B4t working.

will only work if the optimiser picks index join on table 1

> There are some way to do this with a UPDATE statement ?

to do it with an update statement you need no more than one rows in the
from for each row in the target.

easiest non update statement approach is probably to use a plpgsql 
function with a loop.

basically you need to find another way to do it.