Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Usenet -> comp.databases.theory -> Re: Lots of Idiotic Silly Braces?

Re: Lots of Idiotic Silly Braces?

From: paul c <>
Date: Tue, 24 Jul 2007 19:32:10 GMT
Message-ID: <_sspi.1156$_d2.719@pd7urf3no>

Brian Selzer wrote:
> JOIN is a binary operation, and the semantics of PROJECT are such that
> information is always lost, except in the trivial case that involves the
> entire heading.
> ...

So, you are saying that it's okay for binary ops to lose operation but it's also okay for other ops to lose information if that is their "semantics" (which I presume is the same as their meaning). Whereas that's not okay for UNGROUP because either 1) it is not binary and/or 2) its meaning does not allow the loss of information.

As for 1), it's a new on on me that a unary operation must not lose information. Haven't heard that one before, for goodness' sake, why would anybody think that?

As for 2), I've been using D&D's definition, which has a contrary result. What is yours?

> Why is it critical?
> Due to the information principle. If information is lost during an
> UNGROUP, then that means that the set of tuples in an rva is greater than
> the sum of its parts.
> ...

The IP requires only that information be manipulated and presented by its values, it says nothing about which operators are allowed to discard information.

> And also due to the Closed World Assumption:
> ...

The CWA concerns a single relation, not what result a defined operation may produce. It says nothing about when two different relation values are "valid", as you put it. Nor about when the result of an operator is "valid".

Summing up, your arguments seem to be:

1) UNGROUP is not binary, so it is not allowed to discard information.
2) UNGROUP's semantics do not allow it to discard information.
3) The information principle should be applied to the inputs and results 
of UNGROUP but not in the same way to those of other operators. 4) Similarly to 3), the closed-world assumption has an occasional special meaning that applies only to UNGROUP.

And yet, if you work through their definitions, D&D ultimately define UNGROUP in terms of only <NOT>, <REMOVE>, <RENAME>, <AND> and <OR>. It seems that one if one accepts your arguments then there is something wrong with D&D's definition because certain combinations of their operators would cause the IP and the CWA to be violated!

As I said before, I could be content with another formal definition of UNGROUP, but first I would have to see it!

p Received on Tue Jul 24 2007 - 14:32:10 CDT

Original text of this message