Re: cdt glossary 0.1.1 (repost)

From: Brian Selzer <brian_at_selzer-software.com>
Date: Thu, 19 Apr 2007 13:13:07 GMT
Message-ID: <DVJVh.1184$H_.130_at_newssvr21.news.prodigy.net>


"paul c" <toledobythesea_at_oohay.ac> wrote in message news:rJyVh.93528$6m4.2343_at_pd7urf1no...
> Brian Selzer wrote:
>> ...
>> I'm not really sure I understand what you're trying to say here. I would
>> like to see a definition of transaction that is expressed in terms of
>> what is different between successive database states. If you focus on
>> results, some of the information submitted by the user is lost in
>> translation. That's why I get a bad taste in my mouth when I see the
>> imperatives insert, update, and delete translated into assignment.
>> ...
>
> That is a paragraph I think I could have written at one time. Today, I
> think it stands for an attitude that I think is obsessive and technocratic
> (eg. anti-humanist) although I wouldn't go so far as to call it by the
> four-letter word that is usually taken to mean the same thing. I don't
> intend that comment to demean you but I hope you will be able with time to
> explore other perspectives as you obviously have technical ability.
>

Here's another perspective. If you look at an evolutionary graph for a database, you'll find that it is a multigraph, since there is often more than one edge between nodes (consistent database states). If a transaction is defined in terms of a single node, or even in terms of two adjacent nodes, then you can't determine which of the many edges was used to arrive at the current node, and consequently, you can't determine what is different in order to enforce constraints. If, on the other hand, a transaction is defined in terms of a single edge between adjacent nodes, then there's no question about what is different, so there's no problem enforcing constraints.

> p
Received on Thu Apr 19 2007 - 15:13:07 CEST

Original text of this message