Re: About LMT's -- nice article

Following up on Jonathan Lewis, 17 Feb 2003:

>
>
> I think this is probably the first time that someone has
> implied that I might indulge in a cover-up in order to
> promote Oracle.
>

I don't think I implied anything, Jonathan. IIRC I addressed the article, not you. I think I've been around long enough to know you wouldn't do that. Besides, it's not Oracle: it's LMTs. There is (still?) a difference? ;)

>

>>There is, like, nearly four years in-between!  Read metaclick
>>patch bundle bug descriptions if you think they were usable
>>in production before that.


> 

>     In passing - if you've read the 9.2 patch set notes, can I assume

>     that you don't use bitmap indexes, parallel query, or a couple of

>     the other features that have been around since at least 7.3 ?

Hehehe! Parallel query I'll gladly use. A couple (actually, most!) of the other features, as well.

As for bitmap indexes, as soon as anyone can show me any practical case in all the databases I've seen so far where they would be useful, yes, I'll use them. Until then, no.

Yes, I have read your book. And many others. The darn things are so specific it's hard to find a demonstrable case where they'd be useful. In a real life database. Of course a case can be made for their use in a demo environment. So far, they've been next to useless in all cases I've seen for a practical use. Sorry, but that's my experience.

Then again, I've also worked with a guy who routinely creates bitmap indexes on anything that is a varchar2(1). Then he wonders why he gets so many table scans...

But as always, I remain open to new ideas and concepts. As soon as I find them useful, they get used. Not because they are the flavour of the day or just fashionable. That was the point I was trying to make.

-- 
Cheers
Nuno Souto
nsouto_at_optusnet.com.au.nospam