To: pgsql-hackers@postgresql.org
Subject: Re: 7.4 compatibility question
References: <200310231633.h9NGXYk24552@candle.pha.pa.us>
In-Reply-To: <200310231633.h9NGXYk24552@candle.pha.pa.us>
From: Greg Stark <gsstark@mit.edu>
Organization: The Emacs Conspiracy; member since 1992
Date: 24 Oct 2003 00:06:35 -0400
Message-ID: <87brs71dpw.fsf@stark.dyndns.tv>
Lines: 21
User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii

Bruce Momjian <pgman@candle.pha.pa.us> writes:

> Michael Brusser wrote:
>
> > Optimizer improvements
> >  => this tells me nothing. I suppose this could be a minor internal code
> > tweak, which does not affect me. On the other hand this could be a major
> > breakthrough, so now I can run some stupid query which would take
> > a week to complete in the previous release. How do I know? 
> 
> Yes, this is always very hard to explain. The optimizer itself is
> complex, and uses complex terms like merge join and key pruning.  It is
> hard to explain what queries will be affected, though the basic issue is
> that the optimizer will choose a better plan more frequently.

One thing that might be worth mentioning is that "WHERE foo IN (subquery)"
type queries are much improved. That's a one of the more common complaints
about 7.3 and previous and it's one that fairly easy to recognize.

-- 
greg