Re: relational tables and objects

Alfredo Novoa wrote:
>On 31 Oct 2002 23:44:51 +0100, hidders_at_REMOVE.THIS.uia.ua.ac.be (Jan
>Hidders) wrote:
>
>>>The relational model is the best known form for representing and
>>>manipulating data.
>>
>>It is? Gee, I guess we then better give a call to Donald Knuth and tell him
>>to rewrite his "The Art of Computer Programming" books. :-)
>
>It is obvious that I mean it is the best known form for representing and
>manipulating data at logical level. Because the relational model only has
>sense at logical level.

Exactly, and the question was if in this case there was the need for such a logical level. Whether that should be per se the relational model is something I will be happy to debate with you some other time. :-)

>>>The relational model was not designed for hard disks nor RAM. It is a
>>>logical model.
>>
>>Indeed, a logical model for large shared data banks. There was a reason
>>why Codd chose that title.
>
>But you can construct shared data banks with RAM, tapes, punched
>cards, etc.

Of course, but in this case we were not talking about a shared data bank and probably not even a large one. If in addition to that your data is always in memory and there is no ad-hoc access, then I am beginning to have my doubts.

>>>Performance and the relational model are independent. A main memory
>>>relational DBMS may perform as fast as any kind of main memory DBMS.
>>
>>Sure. But the relevant question here is if you need a DBMS, whatever the
>>type. Not every application needs a DBMS.
>
>But you questioned the performance of relational data management
>libraries.

Yes, as compared to data structures specifically designed for the application at hand.

• Jan Hidders