Re: OO versus RDB

On 18 Jun 2006 02:49:00 -0700, David Barrett-Lennard wrote:

Dmitry A. Kazakov wrote:
On 17 Jun 2006 19:51:27 -0700, David Barrett-Lennard wrote:

I find it difficult to respond! Despite all that youv'e said I don't
really understand your concerns with my original post. There seems to
be a lot of generic statements using ill defined terms. ie a lot of
hand waving! I can neither agree nor diasagree with what you are
saying. Rather I keep on asking myself: what point are you trying to
make? Can you please say something more specific - perhaps using a
real example?

Hmm, how could it come to any example when a methodology [of comparison] is
in question?

What methodology?

Yours (:-))

Your point was: RM has simple transparent semantics? [Is it right?]

Yes

My objection/question was: so what? What makes this [any] semantics good
for solving problems? [hint: you have to show correspondence of semantics]

(There are many semantically simple things, toothpaste, for example.)

I'm neither saying OO is always best nor RM is always best. My
original post proposes a criterion for when to prefer one paradigm over
the other.

I only like RM for storing "factual" knowledge about external entities.

Exactly this is flawed. mAmsterdam has already answered that.

In this limited domain, RM seems ideally suited.
It is impressively simple to understand, powerful and flexible. Every
record of every table can be understood in isolation. OO enjoys no
such simple semantics. Please tell me which statements in this
paragraph you disagree with.

Record is a value of some relation. What is there to understand [in
isolation]? The rules of relational algebra? How RA is related to the real
world things?

I'll try to give you an example. Consider a ring of integers. If I say 3,
what is there to understood? Now, some "factual" knowledge in I: 5, 15386,
91, 3. So?

The structure of integers [ring] is the whole semantics of. There is
nothing beyond it. If you use integer to count pennies, you do it at your
own risk. Integers aren't pennies. Same is true for artifacts of any model
you take, be it OO, RM or Peano arithmetic.

I made these claims using the example of representing large amounts of
knowledge about humans. If you disagree with me, why not use the same
example, and point out some advantages of OO?

The advantage of OO is a higher abstraction level. Which would be rather
difficult to formulate, but can be explained on examples. With ADTs, the
parent-child relation can be named as lineage. You can deal with lineages
instead of individual relations. You can express transitivity of
parent-child relation. That it is antisymmetric. You can evaluate a closure
of this relation and paths of the directed graph induced by it [remember
power set?]. You can give a time aspect to the relation, which would be
different from the time span of the program / database. You can replace
Boolean result of the relation with a continuous confidence measure, like
probability, possibility, necessity. You can add mood to the relation,
making it conditional or imperative, etc.

My criterion is : OO should limit itself to things that "are what they
are". That doesn't stop OO from working well in all the above areas.

Please comment on my criterion. If it is flawed then there exists a
counter example. I challenge you to find one!

No, that might be a second step. Because first we have to agree on a
methodology. "What are things?" is not a proper sentence in my language. So
the question is meaningless.

--
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de
.

Relevant Pages

  • Re: OO versus RDB
    ... really understand your concerns with my original post. ... how could it come to any example when a methodology is ... you have to show correspondence of semantics] ... I'm neither saying OO is always best nor RM is always best. ...
    (comp.object)
  • Re: OO versus RDB
    ... really understand your concerns with my original post. ... be a lot of generic statements using ill defined terms. ... RM has simple transparent semantics? ... My criterion is: OO should limit itself to things that "are what they ...
    (comp.object)
  • Re: OO versus RDB
    ... really understand your concerns with my original post. ... RM has simple transparent semantics? ... for solving problems? ... of cases where RM does not suit, as a paradigm of software construction. ...
    (comp.object)
  • Re: One under
    ... >The original post was a thoughtful contribution about an upsetting ... and all the respondents seem able to do is bitch about ... >semantics, when the meaning was perfectly clear all along ...
    (uk.transport.london)