Re: Domain Definition

Thanks Joe,

See comments below...

"--CELKO--" <71062.1056_at_compuserve.com> wrote in message news:c0d87ec0.0301090843.44cbc41a_at_posting.google.com...
> >> Should a domain or collection of values be defined solely based on
> a default
> system representation of values independent of associated meaning, or
> should
> semantics of the values also play a part in determining and partition
> in domains? <<
>
> I would say semantics as a vital part of a domain. Consider
> temperatures; it is not enough to say that the temperature is 25
> degrees without giving the scale (Celsius?, Fahrenheit?, Kelvin?)

Chris Date, in his "DBMS Letter to the Editor" (http://www.dbmsmag.com/dateltr), gave a certain perspective on this issue (in response to yours, I might add :-) ). From what I gather, the major salient points seem to be 1) There is no such thing as a domain of units (e.g. domain of feet or a domain of pounds), there are only domains whose values are measurements in feet or pounds) and 2) a domain whose values are measurements should specify the (single?) unit for all values at the time the domain is specified.

I work on modeling and specifying systems that deal with events concerning parts, each of which have it's own unit of measure in several different contexts (the customer's unit of issue, the supplier's unit of procurement, inventory store uom, etc.). Now, Mr. Date never said proscribed the use of a domain of units_of_measure that can be defined as the entire set of allowable or recognized units of measure. So I take it to mean that it is ok in cases where multiple units of measure might apply across tuples in a relation to either consider attributes representing a quantity or measurement in conjunction with a specified unit of measure so that the quantity could actually be considered a composite attribute.

I've seen other authors express the same sentiments as Mr. Date. For example, Peter Chen, in his seminal paper "The Entity-Relationship Model-Toward a Unified View of Data", defines values sets, which can also be thought of domains also, of 'FEET' and 'INCH' and talks about mapping equivalence across these two domains....(see page 12).

>
> >> In other words, if two domains are equivalent in terms of the set
> of values they contain, but the description or meaning of the values
> of at least one element of both domains differs in terms of semantics,
> should the domains be considered not equivalent? <<
>
> Yes. I can measure the same thing with (x,y) or polar coordinates.
> Both systems will be an ordered pair of real numbers, but with totally
> different rules for computations.

Ok, but now we are getting into the realm of user-defined types with robust operators. Given that SQL DBMS's still do not fully support the concept of type=domain, how do we synthesize a model of domain from limited and stilted representations in today's legacy SQL databases? The impetus for my questions lies in the fact that I desire to specify or model domains from already existing implementations that do not express such semantics.

Thanks again,

Daniel Guntermann Received on Fri Jan 10 2003 - 22:59:35 CET