Re: A pk is *both* a physical and a logical object.

From: Brian Selzer <brian_at_selzer-software.com>
Date: Tue, 31 Jul 2007 17:34:29 GMT
Message-ID: <FoKri.42344\$Um6.17423_at_newssvr12.news.prodigy.net>

"David Cressey" <cressey73_at_verizon.net> wrote in message news:zaFri.3043\$6f4.2362_at_trndny01...
>
> "Brian Selzer" <brian_at_selzer-software.com> wrote in message
> news:kREri.12113\$eY.9117_at_newssvr13.news.prodigy.net...
>>
>> "David Cressey" <cressey73_at_verizon.net> wrote in message
>> news:qOjri.5936\$Lu4.4030_at_trndny03...
>> >
>> > "Brian Selzer" <brian_at_selzer-software.com> wrote in message
>> > news:44eri.25057\$RX.13408_at_newssvr11.news.prodigy.net...
>> >
>> >> I disagree totally. Keys can be the target of an update.
>> >> Propositions
>> > are
>> >> not removed or replaced: they are assigned a different truth value.
> And
>> > it
>> >> is the referents of the propositions that bind them.
>> >
>> > This is pretty much meaningless. Temperatures can be summed. That
>> > doesn't
>> > mean that the result is meaningful. If fact, one sums a collection of
>> > temperatures as an intermediate step in calculating the average
>> > temperature.
>> > While the sum of a collection of temperature has no meaning, the
> average
>> > does.
>> >
>>
>> I don't think temperatures /can/ be summed: a temperature may be
>> identifiable by its magnitude, but that magnitude isn't all that is a
>> temperature. One does not sum temperatures, one sums the measurement or
>> representation of temperatures' magnitudes.
>>
>
> Let me be a little more precise: a collection of numbers can be summed,
> even if each of those numbers represents a temperature. The resulting sum
> is not meaningful, although if one were to then divide the sum by the
> count, the result would be the average temperature, which is meaningful.
>
> This distinction between "summing temperatures" and "summing numbers
> that
> represent temperatures" is at the heart of the running debate between you
> and me (and possibly between you and the newsgroup consensus) concerning
> the update of keys.
>
> Methematically, one can surely update (some of) the key values of (some
> of)
> the rows in a relation, yielding a new relation. The question is what
> this
> represents, in the real world. One case I want to discard is "error
> correction". The relationship between a database that is in error and the
> subject matter it purports to represent is generally confused.
>
>
>
>
>> > So, what does it mean when a key is updated?
>>
>> What does it mean? Well, that depends:
>>
>> (1) It depends upon the number of keys defined on the relation schema and
>> whether all or only some of the keys are the target of the update.
>
> If the before and after values of the update are both correct, how can
> you
> update only some of the keys without breaking one or more of the FDs the
> underly the relation?
>

Because a functional dependency is a state constraint. For example, consider a relation {E, P} that represents an ordered set such that E and P are both keys,
before: ('A','B','C','D','E')

{{E='A', P=1},{E='B', P=2},{E='C', P=3},{E='D', P=4},{E='E', P=5}}

after: ('B','A','C','D','E')

{{E='A', P=2},{E='B', P=1},{E='C', P=3},{E='D', P=4},{E='E', P=5}}

Here a key update occurred, since the positions of 'A' and 'B' in the oset were swapped, yet the functional dependencies E-->P and P-->E are satisfied in both relation values.

>
>
>> (2) It depends upon whether each value for the key permanently identifies
>> /the/ individual referenced by the tuple that contains the key value.
>> Emphasis on /the/ because it is possible that not all key values in a
> tuple
>> refer to the same individual. An update may target a key that is not the
>> subject or the entire subject.
>
> If there are multiple keys in a relation, they are bound together by FDs.
> How can one change one key without either changing all the other keys or
> breaking an FD?
>
>
See above. Received on Tue Jul 31 2007 - 19:34:29 CEST

Original text of this message