NetNews Usenet Archive 1992 #27/NN_1992_27.iso/spool/comp/database/7904

Text File | 1992-11-17 | 2.5 KB | 47 lines

Newsgroups: comp.databases
Path: sparky!uunet!usc!sdd.hp.com!zaphod.mps.ohio-state.edu!darwin.sura.net!jvnc.net!yale.edu!ira.uka.de!math.fu-berlin.de!unidui!rrz.uni-koeln.de!Germany.EU.net!mcsun!chsun!bernina!neptune!nugget.inf.ethz.ch!marti
From: [e-mail] (Robert Marti)
Subject: Re: Normalizing 2NF -> 3NF
Message-ID: <[e-mail]>
Sender: [e-mail] (Mr News)
Nntp-Posting-Host: nugget.inf.ethz.ch
Organization: Dept. Informatik, Swiss Federal Institute of Technology (ETH), Zurich, CH
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Date: Tue, 17 Nov 1992 08:10:19 GMT
Lines: 34

In article <[e-mail]>, [e-mail]
(Niels Elgaard Larsen) writes:
|> [e-mail] (Robert Marti) writes:
|> >[...]  Yhe solution with R'(A*,B) and R"(B*,C) is much better than
|> >the other one with R'(A*,B) and R"(A*,C).  Why?  Well, if you
|> >perceive an FD B -> C in the real world, you probably would
|> >like your system to enforce this integrity constraint. [...]  The
|> >simplest (and in some systems still the _only_) way to enforce an
|> >FD is to make its left hand side a key for a relation consisting of
|> >the attributes listed on both its left and right hand side.
|>
|> Yes, but this might not be efficient. Fx, if there are two groups of users
|> that wants to see an (A,B) and an (A,C) view resp., it requires a lot of
|> unnecessary joins. Eg, (A,B,C) == (Name,Address,Phone#). You typically look up
|> addresses from names and phone-numbers from names---not phone# from addresses.
|>
|> A DBMS should be able to check other integrity constraints than primary keys.

Agreed, they _should_.  But like I said, many (still) don't.  And
since you raise the issue of efficiency, I'd just like to point out
that the R'(A*,B), R"(A*,C) solution requires that you join R' and R"
in order to check the FD B -> C which in turn is much more expensive
than a solution with a unique index.  So it all boils down on the
access patterns.  (I know, you siad this yourself in your earlier
posting.

Finally, I'd be very surprised if the FD Address -> Phone# held indeed
in Denmark.  It certainly doesn't in Switzerland.

--
Robert Marti                    |  Phone:    [phone]
Informationssysteme             |  FAX:      [phone]
ETH-Zentrum                     |  E-Mail:   [e-mail]
CH-8092 Zurich, Switzerland     |