Re: OTLT again
Date: Sat, 13 Nov 2004 08:57:24 -0500
Message-ID: <dZ6dnX1qtZVUjwvcRVn-rg_at_comcast.com>
"Zsolt Ujvari" <reply_at_group.com> wrote in message
news:4195e97e$1_at_news.broadpark.no...
> I searched on the subject and I know/expect to be trashed but I just have
to
> pop it up anyway because I didn't see any discussions on speed
gains/losses
> with a OTLT, only that you loose constraints, data type checks etc.
I think that OTLT comes with a host of problems. But you sound like you have already read up on them. So I'll skip trashing OTLT as a solution, and I'll definitely skip trashing you.
> One lookup looks tempting now, but I'm basically looking for the best
> sollution for my problem...
Instead of discussing the solution with you, I'd like to discuss the problem.
Most of the people who adopt OTLT do so for reasons like the one you
outline. They want to enable user defined attributes. User data is
supposedly different from metadata.
Thus:
Generates metadata.
INSERT (ATTRIBUTE_TYPE, ATTRIBUTE_NAME) INTO LOOKUP_TABLE_MASTER
VALUES (:NEW_ATTRIBUTE_ID, :NEW_ATTRIBUTE_NAME);
Generates user data.
One of these is seen as "better" than the other from several points of view.
If I start talking about OTLT, I'll start trashing, and I said I wouldn't
do that.
But the real discussion ought to be about user defined attributes. Is this
a good idea or a bad idea? If it's a good idea, then we can focus on the
best way of implementing it. If it's a bad idea, we can focus on the best
way of talking the people who set the requirements out of asking for this
feature.
CREATE TABLE NEW_ATTRIBUTE_LOOKUP
(NEW_ATTRIBUTE_ID NUMBER,
NEW_ATTRIBUTE_NAME TEXT);