Re: foundations of relational theory?
Date: Fri, 31 Oct 2003 07:40:02 +0000 (UTC)
Message-ID: <bnt3kh$jqd$1_at_sparta.btinternet.com>
"Bob Badour" <bbadour_at_golden.net> wrote in message news:cd3b3cf.0310300557.12904e81_at_posting.google.com...
>> constraints. For instance, one might have a constraint that an
> attribute in one relation equals a sum of values from another
> relation. Or one might just make the relation a view that derives its
> attribute as the sum.
Okay you've piqued my interest.
Why would you EVER " have a constraint that an attribute in one relation equals a sum of values from another relation."? To do so would be to store redundant information. I would never dream of doing this as it requires two updates for every update. I would just use the standard MV construct of a "symbolic" or correlative. Ultimately the same idea as a derived view - a column that has no value other than that derived by another operation. this column may be treated to all intents and purposes as a physical column.
The crux of my query is the fact that by default I would implement in your latter way as would most of us I believe - so what is it about your theoretical relational database that would encourage people to a sub optimal implementation? Received on Fri Oct 31 2003 - 08:40:02 CET