Re: VAT rate or VAT amount as a column?
Date: Sun, 06 Aug 2006 16:06:26 GMT
Message-ID: <6soBg.35802$pu3.466299_at_ursa-nb00s0.nbnet.nb.ca>
Emily Jones wrote:
> Your suggestion destroys normalisation. > > And who said anything about a trigger? That seems to be your idea. > > Emily
Emily,
You can safely ignore Ken; he speaks mostly to hear his own voice. He doesn't seem to have much interest in theory while he does seem to have great interest in products and product features.
Using a trigger means one must not only trust the dbms, one must trust that the trigger was active and in place at all times in the past. Otherwise, an update could sneak past it.
When derived stored data is truly managed by the dbms, the dbms manages 'freshness' to make sure no stale data sneaks past. Triggers don't do that.
Whether his suggestion destroys normalisation is further complicated by the fact that the products of his interest generally fail to provide sufficient distinction between the physical and the logical and subsequently fail to provide adequante independence between the two.
> "Kenneth Downs" <knode.wants.this_at_see.sigblock> wrote in message > news:c6fhq3-r32.ln1_at_pluto.downsfam.net... >
>>Emily Jones wrote:
>>
>>
>>>A view isn't the same as a base table though is it? Materialised or not.
>>
>>Right.
>>
>>My point is that if we allow system controlled derived data, and we trust
>>the values we are given because we trust the db server, what does it
>>matter
>>if we put the values in a view (materialized or not), or into the base
>>table? Further, putting them into the base table is just plain simpler
>>and
>>more convenient.
>>
>>The trigger is the mechanism that lets us accomplish a system-controlled
>>derived value, and to put it where it is easiest to use: directly into the
>>base table.
>>
>>
>>>"Kenneth Downs" <knode.wants.this_at_see.sigblock> wrote in message
>>>news:q0heq3-ghe.ln1_at_pluto.downsfam.net...
>>>
>>>>Emily Jones wrote:
>>>>
>>>>
>>>>>I thought a table with a derived column in it ISN'T normalised.
>>>>>Functionaly dependent not on the Primary Key, or something. No?
>>>>>
>>>>>
>>>>
>>>>Correct. But neither is the data in many views. Both are system
>>>>controlled
>>>>(if you go triggers) so whatever theory allows views allows the
>>>>materialized columns.
>>>>
>>>>--
>>>>Kenneth Downs
>>>>Secure Data Software, Inc.
>>>>(Ken)nneth_at_(Sec)ure(Dat)a(.com)
>>
>>--
>>Kenneth Downs
>>Secure Data Software, Inc.
>>(Ken)nneth_at_(Sec)ure(Dat)a(.com)
Received on Sun Aug 06 2006 - 18:06:26 CEST