Thanks Mike, good insight. Sounds like shoot from the hip kind of
programming. (good OO developers are NOT shoot from the hip) Probably
because people view web stuff as "having to happen right now" ie we can't
wait to understand the problem, plan, end execute but a more fire, ready,
aim paradigm. If people would think of their web sites as representing the
quality of their business then they might be more ready , aim, fire. For
example, if someone defaced their corporate headquarters so the public could
see it ; someone would be out there right away cleaning the wall. But a
poor web site (application) might be up a lot longer than it takes to clean
that wall. But which is seen by more people? (usually the web site)
Jim
--
Replace part of the email address: kennedy-down_with_spammers_at_attbi.com
with family. Remove the negative part, keep the minus sign. You can figure
it out.
"Mike Sherrill" <MSherrill_at_compuserve.com> wrote in message
news:vpoacv40c8spahe9pm7i6chkd1jc9kt08n_at_4ax.com...
> On 13 May 2003 20:55:30 -0700, wizofoz2k_at_yahoo.com.au (Nuno Souto)
> wrote:
>
> >I have a Java "system architect" on record saying that relational
> >databases are useless for OOD because they don't allow him
> >to build relationships on the fly as needed! He'd much rather
> >have a flat file and build "his own" relationships.
> >
> >Is that loopy or what? I mean, if there is ONE FUNDAMENTAL
> >reason for the use of relational it's PRECISELY because it
> >lets one build relationships as needed. It's the whole FOUNDATION
> >of the darn thing!!!
>
> A SQL database that stays close to the relational model does let you
> build relationships as they're needed. But letting you (a DBA) build
> relationships /as needed/ isn't quite the same thing as letting an OO
> programmer build relationships /on the fly/.
>
> Based on my reading of various database and OO programming newsgroups,
> /some/ OO programmers want to build "relationships" like this on the
> fly. (That is, they want to build tables like this at run time.)
>
> Flight Pilot Seat Chair Passenger Color
> --
> 133 Jones 0 Left <null> blue
> 133 Smith 0 Right <null> blue
> 133 Jones 31 A Rubble red
> 133 Jones 31 B Flintstone brown
>
> What this table means <COUGH> is that Jones is piloting flight 133
> (because he's in the left chair of seat 0); he has a blue security
> badge. Smith is the copilot (right chair of seat 0). Rubble and
> Flintstone are passengers in 31A and 31B. Rubble has red hair, and
> Flintstone has brown hair.
>
> [It's not clear to me whether Rubble's hair was red when he booked the
> ticket, when he had his passport picture taken, when he had his
> driver's license taken, or when he showed up at the gate.]
>
> I can't yet explain why they think things like this must be done at
> run time. I do believe they think flat files are superior to views.
> (You could generate this view from 5NF tables. It would be stupid,
> IMHO, to put security badges and hair color in the same column, but I
> could probably do it if I had to.) But I suspect that they believe
> database designers just take too long to build tables. (That is, that
> we take too long to understand what the data means and how it fits
> together.)
>
> In my own defense, I'll just point out that we (you and I, anyway)
> work with /shared/ databases. You can't rightly share data without
> sharing its meaning.
>
> --
> Mike Sherrill
> Information Management Systems
Received on Fri May 16 2003 - 19:53:52 CDT
