Re: The TransRelational Model: Clarifications
From: Costin Cozianu <c_cozianu_at_hotmail.com>
Date: Mon, 15 Nov 2004 14:55:43 -0800
Message-ID: <2vsqg2F2ogcr2U1_at_uni-berlin.de>
>
>
> Should prove what?
>
> Even if a new proposal is significantly worse by a constant factor on
> the simplest cases, it could be very interesting whether it is a lot
> better on the complex cases where the traditional approaches perform
> horribly.
>
> Most practicioers tend to avoid complex queries due to the performance
> issues. An implementation which makes performance independent to the
> complexity of the queries would be an important advance.
Date: Mon, 15 Nov 2004 14:55:43 -0800
Message-ID: <2vsqg2F2ogcr2U1_at_uni-berlin.de>
Alfredo Novoa wrote:
>> My >>reasoning was that any new proposal for storage organization should prove >>itself on some simple examples before checking it against more complex >>ones.
>
>
> Should prove what?
>
> Even if a new proposal is significantly worse by a constant factor on
> the simplest cases, it could be very interesting whether it is a lot
> better on the complex cases where the traditional approaches perform
> horribly.
>
> Most practicioers tend to avoid complex queries due to the performance
> issues. An implementation which makes performance independent to the
> complexity of the queries would be an important advance.
That would not be only an important advance but also a perpetuum mobile. Received on Mon Nov 15 2004 - 23:55:43 CET