Re: 3vl 2vl and NULL

On 4 Jan 2006 17:40:57 -0800, "dawn" <dawnwolthuis_at_gmail.com> wrote:

>Gene Wirchenko wrote:
>> On 3 Jan 2006 18:40:46 -0800, "dawn" <dawnwolthuis_at_gmail.com> wrote:
>>
>> [snip]
>>
>> >I think of modeling data with both tags (domain names) and values.
>> >Take a simple screen for displaying a person's first name, last name,
>> >all values for e-mail address and first names of all children. We can
>> >surely model this data with multiple relations and we could also use a
>> >single relation with nested relations to model it if we add in data for
>> >ordering, but we cannot model this view of the data with an SQL view.
>> >We could potentially have a cartesian product view that includes all of
>> >the data for the screen, but it does not use the same data model as the
>> >view. The data model for the screen is not in 1st normal form and SQL
>> >requires 1NF.
>>
>> A screen is not a database. It is more like a report. There is
>> no reason for it to be in 1NF.
>
>Correct. But whether I'm modeling data for a user interface, a
>message, or a database, I don't see why I would want to use a different
>data modeling approach if I don't have to. I would like to do

     That is very close to "I have a hammer. Every problem is a nail."

>end-to-end data modeling without 1NF and with a 2VL. And I can, so I'm
>planning to spread the word. There is no reason to go through the

     You can also jump off a bridge and spread the word about that. How do you know you are not doing the analogous here?

>headaches and risks of an impedence mismatch anywhere between a UI or
>message and the database. I don't expect you will tell me I'm right,
>Gene, but I do have a viable approach -- one that works. So, now how
>do we decide who has the better strategy? Cheers! --dawn

     *We* do not. *You* have already decided. Faced with that sort of dogmatism, I tend to wait for the train wreck.

Sincerely,

Gene Wirchenko Received on Thu Jan 05 2006 - 18:06:59 CET