Re: An object-oriented network DBMS from relational DBMS point of view
Date: 10 Mar 2007 08:32:04 -0800
> "Dmitry Shuklin" <shuk..._at_bk.ru> wrote in message
> Note, that the concept of references to the rows is also not new and
> was embodied long ago in such famous RDBMS, as Oracle.
I worry that this is yet another system that focuses on technology instead of the information part of IT, and which doesn't understand what data is (data = given fact = proposition).
Considering two bits of data:
- Aristotle is a human
- A human is mortal
Now that seems fine to me. Two different bits of data, with different structures (and hence different predication). A join also gives me "Aristotle is mortal". All is good, and I certainly don't need any "human object" inheritance - my foreign key indicates that connection for me.
So you want to add 'row references' so that the real world
propositions to be recorded has become.
- Row A says Aristotle is a Row Z
- Row Z says a Human is a mortal.
This would add nothing to the data, it has a superfluous extra addressing mechanism that has to be navigated along, and propositions are no longer directly match those recorded in the real world or how we communicate. This is not good.
I also find the overall conclusion to be back to front. Objects are very constrained subsets of general relationships, not the other way around. Regards, J. Received on Sat Mar 10 2007 - 17:32:04 CET