Re: What predicates the following relation represents

• Post for FREE via your newsreader at post.usenet.com ****

"robert" <gnuoytr_at_rcn.com> wrote in message news:da3c2186.0405070404.21f459ac_at_posting.google.com... > "x" <x-false_at_yahoo.com> wrote in message news:<409a64a2_at_post.usenet.com>...
> > **** Post for FREE via your newsreader at post.usenet.com ****
> >
> >
> > "robert" <gnuoytr_at_rcn.com> wrote in message
> > news:da3c2186.0405060759.5f17fbf9_at_posting.google.com...
> > > pbrazier_at_cosmos-uk.co.uk (Paul) wrote in message
> > news:<51d64140.0404070044.487ccbc6_at_posting.google.com>...
> > > > "Paul Vernon" <paul.vernon_at_ukk.ibmm.comm> wrote in message
> > news:<c4u8li$106q$1_at_gazette.almaden.ibm.com>...
> > >
> > > <snip>
> > > > OK I think you understand what I'm saying but I think some others in
> > > > the thread maybe don't. I'm in agreement with Date that saying
> > > > something twice doesn't make it any truer. What I'm also saying is
> > > > that saying something twice doesn't make it wrong.
> > > <snip>
> > >
> > > of course it makes it wrong; unless, of course, you're treating each
> > > table like a VSAM/COBOL flat file. as soon as you join to this table
> > > (and if you're phobic about joins, find another line of work), you
> > > get wrong results. unless you write explicit code to deal with
> > > duplicates, etc.
> >
> > Could you give a counterexample ?

>
> if you mean to the VSAM approach, just define a PK.  duplicates will
> be prevented.

I meant a counterexample to "saying the same thing twice doesn't make it wrong"
You mentioned something about joining tables with duplicates. If someone ALLOW duplicates in tables but DON'T assign any meaning to the number of duplicates, how can he/she "get wrong results" ? I know there could be some performance problems, but you said "wrong results".
I understood "wrong results" as incorrect results.

About the meaning of "saying the same thing twice": 1) it could mean duplicate rows
2) it could mean duplicate operations applied to the database I don't see how allowing multiple, identical INSERT operations without generating an error would get you "wrong results". Take into considerations that these INSERT operations could be made by many applications or users at different intervals in time, not only by the same application in the same transaction.
If the operation is UPDATE, things changes. Executing the same UPDATE operation twice, could produce "wrong results".

All depends on having the same semantic of data and operations on data for all users of the database.
If you standardize the semantic, you shouldn't get "wrong results".

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

• Usenet.com - The #1 Usenet Newsgroup Service on The Planet! *** http://www.usenet.com Unlimited Download - 19 Seperate Servers - 90,000 groups - Uncensored -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=