Re: Relational query with path expressions

From: David BL <davidbl_at_iinet.net.au>
Date: Mon, 6 Apr 2009 18:37:27 -0700 (PDT)
Message-ID: <58cda02f-22d1-4a80-a346-83b91cacba1f_at_u9g2000pre.googlegroups.com>


On Apr 6, 8:31 pm, "Brian Selzer" <br..._at_selzer-software.com> wrote:
> "David BL" <davi..._at_iinet.net.au> wrote in message
>
> news:edd73d6f-678e-4a1c-bd1f-ac59e06f9bb1_at_s12g2000prc.googlegroups.com...
>
> >I get the impression that a lot of the time, a query (or a large
> > portion of a query) can be thought of as a navigation from attribute
> > to attribute using a sequence of joins. This made me think there may
> > be some merit in thinking of attribute names as global in nature and
> > the user only has to specify the start and end attributes and the DBMS
> > is able to deduce the joins that are required. The potential upshot
> > would be that queries are simplified and the user is freed from the
> > burden of understanding the way the data has been decomposed into
> > separate relations. Just recently I read about the so called
> > Universal Relation which takes this approach. However there are a
> > number of criticisms - such as those described in a paper by Kent.
>
> > I think the biggest problem is that quite often there are multiple
> > useful paths.
>
> I disagree with your line of reasoning because relation names have
> significance. For example, suppose that amongst the myriad relations in the
> manufacturing system for a synthetic rubber plant, two stand out: one that
> specifies the recipies that /can be used/ to produce batches of rubber, and
> another with exactly the same attributes that specifies the recipies that
> /are being used/ to produce batches of rubber. Here it is not just the
> juxtaposition of attributes that conveys information but also the names of
> the relations.

I agree with you and my post took exactly that position by emphasising relation names in the path expressions. Sorry if my post was confusing - I was only stating the idea (of using attribute names to specify "access paths") in order to explain why I was rejecting it!

Note in any case that your argument isn't fool proof because proponents of the UR use very descriptive attribute names that in effect distinguish the alternative relations. However Kent suggests these long winded attribute names can get tiresome in practise. Received on Tue Apr 07 2009 - 03:37:27 CEST

Original text of this message