Re: Separating data, index objects

From: Holger Baer <holger.baer_at_science-computing.de>
Date: Mon, 18 Jul 2005 09:11:57 +0200
Message-ID: <dbfknu$gi$1@news.BelWue.DE>

Jonathan Lewis wrote:
> "Joel Garry" <joel-garry_at_home.com> wrote in message
> news:1121115869.254641.18730_at_z14g2000cwz.googlegroups.com...
>
>

>>I have no doubt it would happen regardless even with a full scan hint,
>>since hints are just suggestions.

>
>
> Just happened to catch the tail-end on this,
> so have no idea where the main thread has
> been going. But this comment caught my eye.
>
> Hints are NOT just suggestions - they are
> orders that must be obeyed if they are legal.
> (Though some newer hints are information -
> but even then the information will be taken
> as unarguably true).
>
> A hint will only be ignored if it is out of
> context or hits a bug. The main reason
> that people think Oracle is ignoring a
> hint is that there is very little information
> that explains exactly what each hint is
> supposed to do.
>

I can only speak for myself, but the main reason that I thought a hint can be ignored by Oracle was that Tom Kyte always emphasized that a hint works by adjusting some internal numbers.

However, he slowly seems to change this: <quote>
it removes certain plans from consideration as well.

A hint "should be taken", if it is possible and sensible to take it </quote>

(from
http://asktom.oracle.com/pls/ask/f?p=4950:8:7262638668833597313::NO::F4950_P8_DISPLAYID,F4950_P8_CRITERIA:8912905298920 )

The original statement always lead me to see a hint as a pure mathematical factor in the cbo equation, now it's clear that is not the case.

Regards,
Holger Received on Mon Jul 18 2005 - 02:11:57 CDT