Re: An object-oriented network DBMS from relational DBMS point of view

From: Alfredo Novoa <alfredono_at_gmail.com>
Date: 20 Mar 2007 07:33:10 -0700
Message-ID: <1174401190.537101.288640_at_l75g2000hse.googlegroups.com>


On 19 mar, 21:21, "Daniel" <danielapar..._at_gmail.com> wrote:

> > > I hope you agree that the concept of an instance of an ADT is
> > > absolutely clear.
>
> > "Instance" is a synonym of "object" and it has the very same
> > fuzziness.
>
> I'm not referring to intuitive notions, the algebraic specification of
> ADT's is well understood, there's a literature on that subject.

As Marshall said, there is a very contradictorial literature. "Instance of an ADT" quoted out of context is absolutely unclear as the Java spec proves.

> If
> you don't have some basic familiarity with that literature and its
> terminology, it's difficult to have a meaningful discussion.

I don't have a lot of respect for that literature and I am more familiar with literature that uses better terminology.

ADT is a rather silly synonym of type. I can not imagine a non abstract type nor a type that is not a data type.

> > values can't be created.
>
> I think this statement belongs to the world of intuitive notions
> rather than algebraic specifications,

It belongs to the world of primitive foundational concepts.

> but let me just ask you two
> questions (and they are questions, I'm not trying to make a point):
>
> Are you saying that it is not meaningful to talk about manufacturing a
> custom value as an aggregate of atomic values such as 1.0, "Jun 1,
> 2007", and 33?

Indeed, it is not meaningful to talk about manufacturing values. Values are abstract entities, you can not touch them with your hands nor with machinery.

> Do you think it is meaningful to talk about storing a value in a
> location in computer memory?

It is meaningful to talk about storing a representation of a value in a location in computer memory.

> > Here we have created a variable (instance) named "point" assigning the
> > "ethernal" value Point(0, 0) to it.
>
> > "point" is an object. Isn't it?
>
> In Java, it would be more accurate to say that point is a variable
> that holds a reference to (the address of) a Point object.

Indeed, in Java it would be more accurate to say that point is a variable
that holds a reference to the address of a Point variable that holds a representation of a Point value .

This is complex enough to create a lot of confusion if we are not very careful.

> > > Maybe you can clarify, is "value" typically taken as a primitive
> > > concept, or can it be defined axiomatically?
>
> > Primitive.
>
> Even so, it might be helpful to enumerate the properties that a thing
> must have to be regarded as value, immutability, etc.

More than helpful, it is very necessary.

> Then we can ask
> whether an instance of an ADT satisfies those properties.

First we should define what an "instance" is.

We are walking on a circle.

Regards
  Alfredo Received on Tue Mar 20 2007 - 15:33:10 CET

Original text of this message