# Re: Examples of SQL anomalies?

Date: Thu, 03 Jul 2008 12:31:34 -0300

Message-ID: <486cf0db$0$4040$9a566e8b_at_news.aliant.net>

> On Jul 3, 7:21 am, Bob Badour <bbad..._at_pei.sympatico.ca> wrote:

*>
*

>>It's not a value. It is, however, a datum.

*>*

> Do you have a definition for "value"? I only ask because

*> you're the sort of person who might. I regularly recall your*

*> (ISO?) definition for data; I have found that very useful.*

*>*

*> My personal definition for "value" is something like a*

*> member of a set, specifically a member of a domain.*

Exactly. A data type comprises two sets: a set of values and a set of operations defined using those values. A value then is simply a member of a set which has operations defined over the set.

> I'm less happy with that as time goes by. There's an

*> interesting mathematical definition for "value" that gave
**> me a bit of a surprise when I first saw it:
**>
**> "The quantity which a function f takes upon
**> application to a given quantity."
**>
**> http://mathworld.wolfram.com/Value.html
**>
**> In other words, value is something that is just a
**> member of the range of a function. Still seems
**> odd.
*

That definition for 'value of a function' is more equivalent to 'return value' in a computational context. The RM doesn't make as much distinction among functions and other relations; they are all relations. Data type support, which is both necessary for the RM and orthogonal to the RM, makes more distinction among functions and other relations.

Please note, though, that the wolfram definition is consistent with the element of a set definition above, and the function f is merely an operation defined using the set of values in its range (possibly among other sets.)

> Do you have a definition for "information"?

What's wrong with the ISO 2382 definition? It's the first definition in the standard. Received on Thu Jul 03 2008 - 17:31:34 CEST