Re: Possible bridges between OO programming proponents and relational model

From: Cimode <cimode_at_hotmail.com>
Date: 5 Jun 2006 05:25:44 -0700
Message-ID: <1149510344.074719.294810_at_j55g2000cwa.googlegroups.com>


Keith H Duggar wrote:
> Cimode wrote :
> > Bob Badour :
> > > What are the two dimensions? Can you name them?
> >
> > For bidimensional RAM's, the 2 dimensions are RowAddress
> > and ColumnAddress For tridimensional RAM's (64bit
> > architectures) the 3 dimensions are Block Adress,
> > RowAddress, ColumnAddress Does that make sense? It was
> > written above.
>
> Suppose we have a 16-bit architecture with 2^16 memory
> cells. How many dimensions is the memory?
>
> Suppose we call the 16-bit tuple "address". Is the physical
> memory now 1-dimensional?
>
> Without changing the physical architechure, suppose we
> mentally partition the 16 bits into two 8-bit tuples and
> call one "hi_reg" and the other "lo_reg". Is the physical
> memory now 2-dimensional or is it still 1-dimensional?
>
> Suppose instead we give names to each of the 16 bits.
> Something like:
>
> 0 : pinky 8 : box
> 1 : bit 9 : crate
> 2 : crumb 10 : pallet
> 3 : chunk 11 : container
> 4 : unit 12 : truck
> 5 : finger 13 : train
> 6 : bar 14 : house
> 7 : pack 15 : big_dog
>
> Is the physical memory now a 16-dimensional hyper-cube?
No. In addresses were utilized exclusively by accessing absolute adresses it can be only monodimensional (What BobBadour refers as linear). But most operations are done through relative calculation of hexadecimal adresses. Which is why representation of the above is incorrect. Below would be closer to reality

Absolute adress/Relative Adress

           ColumnAdress1 ColumnAdress2

RowAdress1>  0/0 : pinky    8/0 : box
RowAdress2>  1/1 : bit      9/1: crate
RowAdress3>  2/2 : crumb   10/2 : pallet
RowAdress4>  3/3 : chunk   11/3 : container
RowAdress5>  4/4 : unit    12/4 : truck
RowAdress6>  5/5 : finger  13/5 : train
RowAdress7>  6/6 : bar     14/6 : house
RowAdress8>  7/7 : pack    15/7 : big_dog

After data pages are loaded into memory and showplan are determined how do you think operations of selection are referencing data in memory?
>
> Since the physical memory is a physical object embedded in
> space-time, why isn't it always 3+1-dimensional? Or maybe
> 10-dimensional?
>
> What point exactly are you trying to make Cimode? This
> seems much ado about nothing, to me. What am I missing?
>
> -- Keith --
Received on Mon Jun 05 2006 - 14:25:44 CEST

Original text of this message