Path: utzoo!mnetor!tmsoft!torsqnt!news-server.csri.toronto.edu!bonnie.concordia.ca!thunder.mcrcim.mcgill.edu!snorkelwacker.mit.edu!usc!sdd.hp.com!elroy.jpl.nasa.gov!decwrl!nsc!voder!blia!nobody
From: nobody@blia.sharebase.com (Nobody at all)
Newsgroups: comp.databases
Subject: Re: Query Optimization (Magic Sets)
Keywords: magic sets
Message-ID: <13511@blia.sharebase.com>
Date: 18 Feb 91 20:59:37 GMT
References: <1991Feb14.192012.6084@doe.utoronto.ca>
Reply-To: mike@woodstock.UUCP (Mike Ubell)
Organization: ShareBase
Lines: 25

In article <1991Feb14.192012.6084@doe.utoronto.ca> peter@doe.utoronto.ca (Peter Mielke) writes:
>
>I was just at a talk on relational database query optimization using
>the technique of "Magic Sets".  I was wondering what RDBMSes use any
>form of query optimization (i myself have been bitten quite a number
>of times by not writing my SQL query the right way)? Is it still on
>the user of SQL to format the query in the right way to get the answer
>the fastest way?
>-- 
>Peter Mielke                                    peter@doe.utoronto.ca
>Dictionary of Old English Project               utgpu!utzoo!utdoe!peter
>University of Toronto

Just for the record:  ShareBase I and ShareBase II use an optimiser
that is good for up to about a 4 way join and sometimes to ok on larger
queries.  They are not dependant on the way the query is stated
(ordering of tables and/or qualification clauses (but certain SQL
queries can be stated in a variety very different ways so these might
effect the optimizer).  ShareBase III (in beta test, scheduled for
release around June) does a complete optimization search for up to
about 10 way joins and a pruned search up to 32 way.  As far as I know
it will generate the same query plan for any semanticly equivalent SQL
statements.
(Both optimizers accept explicit directives from the user to set the
query plan if the optimizer does not generate an acceptible one.)