Re: Hibernate and Oracle

On 2/10/07, Robert Freeman wrote:
(SNIP)
> We changed our approach to modeling by allowing them to design their classes
> and generate database DDL based on those classes. I'll admit that I was not
> 100% happy with that change, and the DDL generated had a lot of problems and
> we made a lot of changes (mostly naming, adding FK's, NULL requirements,
> indexes and so on). We found that model was not far from what we modeled at
> the outset, but it really seemed to eliminate some of the initial
> object/database confusion that follows Java to Physical design around. The
> Jury is still out on this approach as far as I'm concerned, but I have been
> moved to believe that it's not a stupid thing.

Well, if you work with good Java "architects", that can produce a sound design (simple and intuitive), for sure that design, once persisted as a bunch of tables, will be sound (simple and intuitive) as well.

But persistence has to be considered right from the start of the design phase. For example, many Java people would be tempted to design a Customer as a class containing NameSurname, Address, IdentificationCredentials, the latter containing one of Passport, IdCard, LicenseToDrive, instead of a simple flat class containing only numbers and strings ... once persisted, each class will map to a table, with obvious performance implications; even if the design is reasonably simple and intuitive.

It takes a Java guy versed in Oracle, or a design team composed of (collaborating) Java and Oracle subteams, to avoid this and similar pitfalls. But then the team will be using traditional relational design techniques essentially ... for both the tables and the Java classes.

Thanks for your chiming in :)
Alberto

-- 
Alberto Dell'Era
"Per aspera ad astra"
--
http://www.freelists.org/webpage/oracle-l