Re: Possible bridges between OO programming proponents and relational model

From: Cimode <cimode_at_hotmail.com>
Date: 3 Jun 2006 10:46:28 -0700
Message-ID: <1149356788.587019.63550_at_f6g2000cwb.googlegroups.com>


<< When you snip as much as you do, it is harder for me to  have a dialog.>>
Communication through forums is not easy I convene. I am not accustomed to exchanges and jargons through NG.

I do not know what *snipping* exactly means. I have however been very clear into clarifying my position whether on a logical and physical layer. The fact of simultaneously considering both into argumentation does not mean I confuse them as you tend to imply.

<< In the RM, and in SQL, the logical model is what it is:  multidimensional. This is true regardless of the  physical implementation.>>
I understand now the source of confusion in communication. I have made clear that I wished this thread would be about implementation but refering to logical requirements seems to be confusing.

I have never denied that RM or SQL are multidimensional. I just do not believe that current implementation of SQL are. And I have not heard any valid argument proving the opposite.

<< It might be easier if you picked one level, physical or  logical, and stuck to discussing that. When you  switch back and forth so quickly, our communication  seems to break down.>>
Thank you for this advice. I understand its wisdom. I will try to be more cautious.

<< You keep saying that physical representation is two  dimensional. But that's not correct. Physical memory  is one dimensional; it is a single line of memory from  address 0 to address n.>>
You are wrong. Most RAM components are two and three dimensional. (adress, columns and sometime blocks for 64 Kbit architecture)

<< If by "bidimensional" you mean that SQL is always  implemented as a simple row store, this is not correct.  There are products that are column stores. There are  other hybrid approaches. It is not so simple.>> No by bidimensional I mean bidimensional in their adressing scheme (row vs columns). The only difference between simple row tore and column store is that they switch dimensions. Their adressing scheme remains logically bidimensional.

Which SQL DBMS products do you refer as being hybrid? column stored and simple row store?
I suggest you check the following link from a friend of mine (Harry McKame) who participated in the initial implementation of System R. He has developed his own version of SQL because he understood the original limitations of current implementations.

http://www.armadillo.fr/english/whitepapers/WHITEPAPER_2004.htm

Thank you for your insight.

Marshall wrote:
> Cimode wrote:
> > <<Okay; you're discussing the physical layer here. (Although your
> > parenthetical remark is incorrect; the choice of a particular
> > implementation strategy isn't what determines whether
> > you have independence or not.)>>
> >
> > You consider relvar and R-Table physical layer? Can't agree with that
> > (must have missed something).

>

> When you snip as much as you do, it is harder for me to
> have a dialog. You talked about physical concerns, which
> I pointed out, and now you say you weren't discussing
> physical concerns, but you've snipped the paragraph
> I was responding to, which discussed implementation
> and physical concerns.
>

> I'll try to be very clear:
>

> In the RM, and in SQL, the logical model is what it is:
> multidimensional. This is true regardless of the
> physical implementation.
>

> It might be easier if you picked one level, physical or
> logical, and stuck to discussing that. When you
> switch back and forth so quickly, our communication
> seems to break down.
>
>

> > << No, I don't think that's correct at all. Physical memory is
> > unidimensional;>>
> > Please explain your point.
>

> You keep saying that physical representation is two
> dimensional. But that's not correct. Physical memory
> is one dimensional; it is a single line of memory from
> address 0 to address n.
>

> If by "bidimensional" you mean that SQL is always
> implemented as a simple row store, this is not correct.
> There are products that are column stores. There are
> other hybrid approaches. It is not so simple.
>
>

> > << Well, I wouldn't do that if I were you. I think your best bet is to
> > study the existing literature on relational implementation.
> > Read ten papers and see if you think OO has something
> > useful to say.>> Thanks for the word of caution.(I appreciate) but I
> > already started doing that. Diversifying sources helps too. ;)
>
>
> Marshall
Received on Sat Jun 03 2006 - 19:46:28 CEST

Original text of this message