Re: Persistence, EJBs and OO-Relational mapping

> from the way you are choosing your arguments, I get the feeling that
 you are
> approaching the problem the wrong way around. "Relational arithmetic"
 is
> nothing natural. It was developed from bookkeeping data in
 spreadsheets.
> That is by no means a very effective look at business rules.

I was talking about business data, not business rules. What's better than Relational Arithmatic to deal with the "bookkeeping data in the spreadsheets" that gets us books from Amazon.com to the right address billing the right amount?

> Reality consists of objects.

??

> Even simple data can always be viewed as an object:
> - data is generated by some method whatsoever

??

> - rules can be applied for relationships between data objects

Could you please give some examples?

> - something is done with the data to produce a result or output
>
> This is why OO-languages add methods to objects.
 

> All the nice "features" of RDBMS can also be expressed in OO-methods.
> Within the near future we will see more object databases with method
> frameworks (like EJB), that add the same features (cascaded delete,
> "referential integrity", constraints, callbacks (triggers)) to their
> products.

My original question was about persistent middletier components, such as EJBs, and their suitability to persist pure relational data. My customers do not work with ODBMS, so can not take much comfort from them.

> Viewing data as objects makes life easier for the programmer.

Depends on who the programmer is, and what programming he/she does. In my view, nothing is simpler than seeing data in its natural format. For some types of data, such as spreadsheets, tables and records are the most natural format.

> Storing object data to relational databases just does not work because
> paradigms do not match.

Isn't the other way true too?

>
> The OO-paradigm is by far more natural and easier to handle than
 relational
> set-arithmetic.

I do a lot of programming in both paradigms. I totally disagree with that assertion.

> We will see a replacement in the next (x) years.
>
> (working on x = 5)
> (x = 20 is more realistic)
>
> Regards,
> Carl

Cheers.

Sent via Deja.com
http://www.deja.com/ Received on Thu Dec 14 2000 - 19:03:54 CET