Re: Fraud Number 3: U-Gene
Date: 18 Jun 2006 09:11:00 -0700
Tony D wrote:
> Cimode wrote:
> > This ignorant trully believes that relations = relvalues and asks
> > whether these values have different types thanks to the previous
> > assumption if the relation have different schemas??
> Now, I did read this thread, and I really like to sort this out. Can
> you describe, without any bluster or invective, what is meant to you by
> the terms "relvar", "relation" and "relvalue" ? You said that "I am
> more a follower of the FP, McGoveran approach who advocate a tighter
> commitment to terminilogy (sic) ..." (16th June, 2:27pm). Since you
> then went on to say that a relation is both the variable and a value (I
> think : what you said was "A relation is BOTH a relvar which represent
> the abstract structure of the relvar and the relvalues which represents
> its matter at a specific point in time.") I'd like an explicit
> description of your view of these three terms, because, like others on
> this thread I've so far taken the CJ Date line on this (even if it can
> lead to overly structural thinking about relations). If it's in one of
> Fabian Pascal's papers, by all means note which one and I'll spend the
> $10 to read it.
> In addition to that, I think U-gene isn't a native English speaker
> either (I think he's Russian ?), so given that a non-native speaker is
> arguing with another non-native speaker, the potential for confusion
//A relation is BOTH a relvar which represent the abstract structure of
the relvar and the relvalues which represents its matter at a specific
point in time//
...I realize I made a TYPO mistake
as it should read..
//A relation is BOTH a relvar which represent the abstract structure of the relation and the relvalues which represents its matter at a specific point in time//
// I'd like an explicit description of your view of these three terms//
First of all, the mathematical aspect of a relation.
One could define a mathematical relation as an absract construct that has purpose of representing and operating a specific time-dependent *combination* of values drawn from arbitrarily defined ensembles of values. In short, a relation can be viewed as a specific ensemblist-combinatory function.
--> variables are defined as elementary *placeholders* for values
(variable is therefore a time independent concept)
--> values are defined as the elementary occurrence (or placefiller) of
a variable in a specifc point in time (value is therefore a time bound
--> the function as a whole defines a more new more complex ensemble of values.
By using predicate logic and characteristics of ensemblist calculus, the whole genius of Tedd Codd was to apply the 3 above characteristics into bridging the gap between informal representation of meaning at human level and formal representation of logical data at mechanized computing level. As a result of his work was defined RM.
--> in RM, mathematical variables were refered to as relation-variable
known also as relvar
--> in RM, Codd tried to apply the concept of mathematical values elementary occurrence as being a specific datum. --> By using the concept of domains (representing ensemble of values of arbitrary complexity), he established that a relation defines a new domain of its own.
--> To define the logical operations and characteristics of relvars in RM, Codd used logical tables (refered as R-Tables) as a possible projection of a relvar in time.
--> In R-tables, he defined the rules by which values (sometime called relvalues) could be drawn from and possibly operated from a specific domain. THe unique combination of these rules constitute a *data type*. By using R-Tables, he associated the representation of the relvar in a column\row logical construct where columns are defined as attributes and rows (respecting predicate theory prerequisites) represent *tuples*.
Knowing this should avoid further confusion:
--> relations, relvars, relvalues are on different level of definition.
They certainly are not to be equated.
--> attribute data type directly refers to formal definition and *possible* operations that can carry on single domain drawn *permissible* values.
--> relvar type refers to the rule that defines the new domain induced by definition of attributes data types. Therefore, relvar type is not the same as attribute data type.
--> R-tables are ONE possible representation of relvars. --> Considering they do not support sufficiently domain characteritics, SQL Tables are not R-Tables.
Hope this clarified. Received on Sun Jun 18 2006 - 18:11:00 CEST