Re: Mixing OO and DB

From: David Cressey <cressey73_at_verizon.net>
Date: Sat, 08 Mar 2008 13:45:54 GMT
Message-ID: <mMwAj.1002$ox3.524_at_trndny04>


"mAsterdam" <mAsterdam_at_vrijdag.org> wrote in message news:47d24ac7$0$14350$e4fe514c_at_news.xs4all.nl...
> David Cressey wrote:
>
> > Given that so many people want an OODBMS, why has one not been built?
> > I think it's because the term OODBMS contains a contradiction in terms.
> > But I'm open to other opinions.
>
> Unfortunately the ones holding those other opinions appearantly
> aren't open to yours - yet?
>
> When I first heard about OODBMS I thought: when will those marketing
> guys ever stop mixing up terms they don't understand?
> Objects /behave/. Data /is/. So I hoped that the 'OODBMS' vendors's
> designers would discover that to come even just a little closer to
> their marketing idiots claims they would have to device a common
> type system (CTS, common to the DB columns and the application
> structs).
>
> Objects behave. When the context of the running program has gone,
> all objects have died / they have stopped behaving / ceased to be
> objects. In order to reincarnate them at a later point in time,
> the application must make sure all data, needed for the rebirth,
> are stored somewhere /outside/ the object system before the program
> stops running. That is /outside/ the part of the system where
> thinking in objects (a.k.a. Object Orientation, OO) makes any sense.
> Without thinking outside the object box you cannot make objects
> capable of reviving.
>

You've expressed my own view better than I could. Thanks.

What gets stored in a database is data. "Objects" can be stored in a database only in suspended animation. The process of reconstituting an object so that it not only remembers its prior state, but also recovers its prior behavior, is by no means trivial.

In the responses to my original comment, lots of people told me about several OODBMS products. They didn't explain why those products failed to catch on. I don't necessarily subscribe to the argument by market share, but before I invest intellectual energy in something out of the mainstream, I need to come to believe that it's likely to be worthwhile.

I think I understand why relational fans don't like OODBMSes. They solve a problem that doesn't have to exist.

I think I understand why OO fans don't like OODBMSes. They expose a layer that the object paradigm abstracts away from their field of view. Received on Sat Mar 08 2008 - 14:45:54 CET

Original text of this message