Re: Problem with indexes

"Niall Litchfield" <n-litchfield_at_audit-commission.gov.uk> schrieb im Newsbeitrag news:3fa795e7$0$255$ed9e5944_at_reading.news.pipex.net...
> "Jan Gelbrich" <j_gelbrich_at_westfalen-blatt.de> wrote in message
> news:bo81h8$1b1m09$1_at_ID-152732.news.uni-berlin.de...
<snip...>
> > DBMS_STATS has been confirmed to be buggy.
>
> I have had a look at this thread and I'm not at all clear that you can
apply
> it to the situation in this case.

Niall, thanx for the comment, but I would not know why DBMS_STATS would not be as
applicable as ANALYZE. The doco does not point out any difference about them.
So wyh should some type of statistics be better or worse for each of them ... sorry, I am just curious.

To be honest, I just realized *later* that the ANALYZE ... DELETE STATISTICS did what I want and statspack did not, for some reason. My first problem was to make my query *run* again, anyhow, and that was done.
And other people with *much* more knowledge than me explained why, and as far as I read their online and offline responses, they *did* find a general problem
with statspack, and they made more sophisticated testing than me.

>We don't know if there were stats or not
> to start with, nor if they are accurate or if there is some other issue
with
> them. Selectively deleting stats might be helpful in specific cases but in
> general it is a poor tactic

True, it makes only sense in that way that it *did* work for my very special case - surprizingly,
and so it *may* work in other cases too, who knows ? If something is stupid but works, it is not stupid (Murphy) - it remains only unclear why unless You meet somebody who knows - here for instance ...
A rebuild would have been even more "stupid", so I went for the "least stupid" solution.

greetings
Jan

Socrates knew that he knew nothing - but how then could he know that ... Received on Tue Nov 04 2003 - 07:09:23 CST