Re: storing survey answers of different data types

On Sun, 26 Apr 2009 10:16:13 +1000, paul c <toledobythesea_at_oohay.ac> wrote:

> Joe Thurbon wrote:
>> On Sun, 26 Apr 2009 02:43:25 +1000, paul c <toledobythesea_at_oohay.ac>
>> wrote:
>>
>>> Joe Thurbon wrote:
> ...
>> Yep.
>>
>>> assuming we are not talking about what Codd called a 'dictionary'.
>> I'm not sure what you mean.
>> ...
>
> Oh, I just threw that qualification in to cover my lack of interest in
> 'dictionaries', as I found his use of that term a little opaque, admit I
> didn't read much of what he wrote about that. For one thing he was
> quite interested in a dictionary that would store and categorize
> programs as well as invoke them (I know this first-hand, although I was
> too shy to ask him about it). One source that mentions it is his 1990
> book, which used to be downloadable for free from acm, maybe still is.
> If not, and if you don't have it, I could try emailing it to you but
> it's pretty big and might not be worth the trouble just for that topic.

Thanks for the offer, but if it's still free I'll get it. If it's not, I rather pay for it.

> I think his 1970 paper hit the nail on the head (without resort to a
> wrench), in similar fashion to Cimode's example, when he described how
> to take a hierarchy apart. Maybe he didn't intend it this way, but part
> of the message seemed to me to be that he was unravelling types as he
> went along, adding attributes as he went. But what seems to me to be
> the most important point in all of this is how being presented with a
> certain data design can change how one thinks about possible application
> requirements, which Cimode illustrated by avoiding 'one table per
> questionnaire' (I doubt that Roy H would object to that).

Yes, I doubt he (Roy) would. I've ended up down a similar path before as requirements came in to
- be explicit in what types were permissible answers, - be able to automatically format presentation of the questions to the user,
- automatically agregate scores across questions, etc

> I think it is very important because there is a point of view that seems
> helpful to must start with when making relational data designs and even
> though I find it hard to express in words, without that attitude most of
> the technical aspects such as SQL or even the algebra don't get one very
> far. As far as I'm concerned some of the terse responses we see lately
> here regarding design questions are too terse, those ripostes should be
> reserved for obvious technical blunders. (This coming from somebody who
> would agree that most of the western world prizes political correctness
> more than correctness.)

As someone who asks design questions here, I don't mind the terseness of some responses. It puts the burden on me to unpack the response, which often leads me to think about the assumptions I am bringing with me. Mind you, the terseness needs to be precise. That's a capbility I lack, but a couple of the people here often display. (Which means I get to practice that, too.)

Cheers,
Joe Received on Sun Apr 26 2009 - 02:53:52 CEST