Re: -- insert instead of an update

"Galen Boyer" <galenboyer_at_hotpop.com> a écrit dans le message news: u3cc7aooz.fsf_at_standardandpoors.com...
> On Thu, 20 Nov 2003, rtd45Rd_at_nospam.org wrote:
> >
> >> Obviously they don't want to live in the same mess as you seem
> >> to be desiring to get in.
> >
> > Hope this was intended to be ironical :-/ My only wish is to
> > discover what's new with that version because some companies
> > are using it so I don't want to close myself some job
> > opportunities. But maybe a Senior Oracle DBA has some good
> > advice to give me, hasn't you, Sybrand ;)
>
> Sybrand's answer wasn't ironical. He just isn't usually all that
> warm and fuzzy.
>
> I try not to criticize designs that I know nothing about, but I
> am suspicious of your design. Why should an update ever cause an
> insert to the same table? I am assuming auditing is stored in
> your entity table? What's your primary key to this entity table?
>
> --
> Galen Boyer

Hi Galen,

The purpose of my design was simply to find a use for instead of triggers. Another purpose was to see how far it was possible to separate the logical view of the data
from their implementation in a table.

Logical view behaves simply as we expect (delete deletes record, update updates and insert inserts)
the underlying table behaves another way (delete marks a record as deleted, update updates but insert the old version into a new row)

My goal is to keep many versions of reference data in my database so I'm testing different solutions (triggers was one, I keep on searching ;-). Needless to say that I give up the "instead of triggers" solution :))

Cheers,
Alkos Received on Mon Dec 01 2003 - 03:48:15 CST