Re: more on delete from join
Date: Wed, 02 Sep 2009 18:50:26 GMT
Message-ID: <S7znm.43523$PH1.39323_at_edtnps82>
Bob Badour wrote:
> paul c wrote:
>
>> Marshall wrote:
>>
>>> On Sep 2, 2:44 am, "Joe Thurbon" <use..._at_thurbon.com> wrote:
>>>
>>>> So, to update a view, rather than update the conclusion directly,
>>>> one must update (one of) the base relvars that are used to derive
>>>> the conclusion/view. That is abductive.
>>>
>>> I guess I see what you mean, but strictly speaking it doesn't seem
>>> to fit the definition. In abductive reasoning, one knows a -> b, and
>>> b, but one doesn't know that it was in fact a that implied b. Whereas
>>> with a view, we do know.
>>> ...
>>
>> We know "a implied b" because we chose to record it. We could choose
>> to record same for a so-called "base relation", so how are they
>> different? Even "base" updates must deal with constraints. As far as
>> a database is concerned, If there is a difference, it is that the
>> recording of one value is physically optimized compared to the other,
>> ie., the difference is physical, not logical.
> > Even then, the ability to store data in ways that differ significantly > from one's base relations is highly desired. (Physical independence)
Exactly. People who want to treat views differently should play tthe ball where it lies and admit that a named view is a physical optimization device (which has been confusing people for many years now). If they don't want views to represent the same relations bases do, they should prevent views from being named, instead of making bald assertions about, eg., implications and conclusions. Any implication the typical algebra allows to be recorded is in fact a proposition. How about giving us some assertions that have got hair on them? Received on Wed Sep 02 2009 - 20:50:26 CEST