Re: Teach SELECT DISTINCT first!

From: Mikito Harakiri <mikharakiri_at_iahu.com>
Date: Tue, 27 Apr 2004 11:10:00 -0700
Message-ID: <awxjc.40$BD6.607_at_news.oracle.com>


"Paul" <paul_at_test.com> wrote in message news:DGwjc.35459$h44.5223395_at_stones.force9.net...
> There's no way in general an optimizer can tell whether or not there are
> duplicates without doing the calculation. Even if f were a relatively
> simple function it would be too much to expect that a relational engine
> should have knowledge of maths built-in.

CAS usually assumed to have this kind of knowledge, so why (in principle) relational engine couldn't have it? The technical difference might be that everything in CAS is an expression, while in SQL we allow function programmed in 3GL language. In foreseeable future no relational optimizer is expected to look inside procedural code, of course. (Note an interesting analogy: it is easy to reason about a function defined decalratively as an expression, and its very hard to reason about fuction defined procedurally.) Received on Tue Apr 27 2004 - 20:10:00 CEST

Original text of this message