Re: A question for Mr. Celko
From: Jan Hidders <jan.hidders_at_REMOVETHIS.pandora.be>
Date: Sun, 18 Jul 2004 09:59:33 GMT
Message-ID: <pan.2004.07.18.10.00.10.517654_at_REMOVETHIS.pandora.be>
>
> Not only would I agree, but I would argue that not allowing them, as
> is the case in today's systems makes things harder in general. The
> system must handle scalar values differently from all other types of
> values, causing special-case handling in the query processor.
Date: Sun, 18 Jul 2004 09:59:33 GMT
Message-ID: <pan.2004.07.18.10.00.10.517654_at_REMOVETHIS.pandora.be>
On Sat, 17 Jul 2004 23:29:03 -0700, John Jacob wrote:
>> I hypothesize that you can have list-valued-attributes inline (as >> is already often the case with strings) and relation-valued attributes >> implemented as separate relations, and still keep things quite simple. >> I'm still investigating this, though.
>
> Not only would I agree, but I would argue that not allowing them, as
> is the case in today's systems makes things harder in general. The
> system must handle scalar values differently from all other types of
> values, causing special-case handling in the query processor.
Not all interpretations of 1NF say that they are disallowed. The one that makes the most sense to me says that the value should be atomic from the database's perspective. In that case you can have any value in an attrbute but you are not allowed to "break it open" which means you cannot have the nest and unnest operations from the nested relational algebra.
- Jan Hidders