Re: Database design

After takin a swig o' Arrakan spice grog, Mark Johnson <102334.12_at_compuserve.com> belched out:
> Gene Wirchenko <genew_at_ucantrade.com.NOTHERE> wrote:
>
>>On Wed, 22 Feb 2006 11:41:05 -0800, Mark Johnson
>><102334.12_at_compuserve.com> wrote:
>>>"Marshall Spight" <marshall.spight_at_gmail.com> wrote:
>>>>Mark Johnson wrote:
>>>>> "Marshall Spight" <marshall.spight_at_gmail.com> wrote:
>
>>>Let's say you have a roster of US Presidents. Surely this is stored in
>>>some database, somewhere.
>
>>>It is entered row by row, tuple by tuple if you will. And their
>>>position is entered, as well. While I can understand that one might
>>>say the ranking would not necessarily apply, it also might. There is
>>>an intrinsic order to the roster, after all. But by definition, that
>
>> *An* intrinsic order?
>
>> Is that intrinsic order by birthdate?
>
> Please. Their position on the roster. JFK, it must be pointed out, was
> not President before the reign of FDR. FDR preceded him, instead.
>
>> Is that intrinsic order the order in which they served? (If so,
>>what about Grover Cleveland?)
>
> Why parenthetical? And why would he not be listed? Is it that you
> don't know if the RM, or whatever you have in mind, could handle the
> scalars - 22, and 24? Could be a roster of local mayors, councilmen,
> what have you.

He is a perfect characteristic case for the problem that the ordering of presidents is an ambiguous matter.

Question: How do you handle the relative ordering of Grover Cleveland and Benjamin Harrison's presidencies?

Cleveland served BOTH before and after Harrison, which means that there is NOT a clear order on a by-president basis.

After all, it is true to say that:

1. Cleveland served before Harrison, and
2. Cleveland served after Harrison.

An attempt to order *by president* breaks down in general, and here, in specific, due to the fact that terms are not forcibly required to be consecutive.

>>>relation can NEVER be sorted?
>
>> The *data* in the relation can be sorted.
>
> Now wait a minute. Obviously - yes. But this theory is supposed to
> hold that a relation cannot be sorted, and cannot store the data in
> proper order.

According to the theory, relations are unordered sets of facts.

Your need for "proper order" is a fabrication in your own mind.

> If the 'data' in a relation is sorted, if each tuple occupies a list
> position with regard to other tuples, then you must admit that the
> relation is ordered. Maybe you want to say it's not set. Maybe it
> never was?

>>>So a set cannot be ordered because to place it in any order is to
>>>redefine it as non-set? So a roster can never be a set and a roster.
>>>To become a set, the most important attribute of that set must be
>>>discarded?
>
>> What attribute is that?
>
> His position in the lineup. And any changes during the game must also
> be approved. That attribute. The sort. The proper order.

You are assuming there to be a unique position; in the case of Cleveland's presidential service, there isn't one. There isn't a "proper order" because the attempt to force these facts into a "proper order" requires breaking the facts.

>> Why do you consider the order to be the most important
>> attribute?
>
> Game's got rules.

Ordering is not one of the necessary rules. It may be useful for some purposes, but it is not necessary to the relational model.

-- 
"cbbrowne","_at_","acm.org"
http://cbbrowne.com/info/wp.html
"(Windows NT) version 5.0 will build on a proven system architecture
and incorporate tens of thousands of bug fixes from version 4.0."
-- <http://www.microsoft.com/y2k.asp?A=7&B=5>