Re: data & code
Date: Thu, 27 May 2004 23:12:34 +0200
Message-ID: <40b659c1$0$561$e4fe514c_at_news.xs4all.nl>
Dawn M. Wolthuis wrote:
> mountain man wrote:
>> x wrote:
>>> Alfredo Novoa wrote:
>>>>Dawn M. Wolthuis wrote:
>>>>>Very good. So, mountain man's "Organizational Intelligence"
>>>>>is the set of all data?
>>>>It is the set of all the bytes in the hard disk :-)
>>>Or perhaps the bits in all devices, wires, media, brains, etc.
>>>in the past, present and future ? :-)
>>The theory is restricted to "computerized OI" but you raise an
>>interesting issue introducing time. You see, the current theory
>>assumes the relationship between truths in set theory (RDBMS
>>storage of data) and the truths operationally engineered between
>>this data and the external world (via a user interface).
>>
>>In time, the world changes, and without doing anything, the
>>data at the heart of any implementation instance of the relational
>>model, loses its integrity, because there is no dynamic built into
>>the fundamental (RM) model.
>
> This is a bit of a tangent, but quite related to your thought above -- I was
> just going back to what I thought were the basics a few decades ago,
> including use of Master files, Transaction files, History files, and others.
> In any post-transaction logs or history files, data from the Master was
> often DUPLICATED on purpose. This makes complete sense in that the Master
> has information at that point in time and can change and we want to have a
> record of the actual transaction as it took place. I think some of this is
> lost on newbies who are blindly applying normalization techniques, not
> recognizing that data in "Master Profiles" (to update the term) is dynamic,
> while data related to "Transaction Documents" must duplicate that data at
> that point in time in order to preserve it for analysis and accurate
> historical information.
>
> Is there anything within relational theory that addresses this other than
> hoping that the designer will recognize the difference between the term
> "Marital Status" in the Master files and the same if important information
> related to a transaction? Is there even any way to partition types of
> relations into, perhaps updated versions of, Master, Transaction, History,
> ...? --dawn
At
http://www.programming-reviews.com/Temporal_Data_and_the_Relational_Model_1558608559.html
are some reviews of what I think is an excellent read on this topic.
Received on Thu May 27 2004 - 23:12:34 CEST