Re: Principle of Orthogonal Design
Date: Fri, 18 Jan 2008 03:13:55 GMT
Message-ID: <TPUjj.93618$EA5.32998_at_pd7urf2no>
JOG wrote:
> On Jan 17, 5:08 pm, Jan Hidders <hidd..._at_gmail.com> wrote:
...
>> Apologies for being lazy, but could someone explain to me in a >> nutshell why this should be possible at all? At first sight this looks >> like complete nonsense to me.
>
> Well, as far as I gather, being able to determine a unique predicate
> for any proposition being inserted into the database is desirable in
> order to allow view updates to be more easily be translated to changes
> in underlying base relations.
> ...
POOD seems too-little discussed to me. Maybe POFN too.
I'm glad the example was about machines because I like the mechanical view, saves a lot of time when I've forgotten to charge the batteries for my mystic-meter. In an ideal world where every design decision was considered against every attribute/role name, I think the example would be considered a straw man - if it were up to me, I wouldn't allow those two tables/relvars to use the same attribute names.
I realize there are people here who might say what's wrong with same role names because the RM doesn't forbid them. All that proves is that there are naive people here as well as elsewhere! I was just laughing over one of CJ Date's latest examples about the tuple that states that the supplier is not in Paris. I think one must cater one's design towards the mechanical advantage a "machine" offers and using the same names seems a bit thoughtless to me.
It might look neat for defining terse union views without having to resort to some kind of 'rename' macro operator baggage but I have a feeling avoiding a rename is too cute and will bite one back, sooner or later. Maybe rename also gives some kind of legitimacy for inserting to both 'sides' of a union view? Received on Fri Jan 18 2008 - 04:13:55 CET