Re: Object oriented database

On Nov 1, 3:57 pm, "Walter Mitty" <wami..._at_verizon.net> wrote:
> <patrick..._at_yahoo.com> wrote in message
>
> news:52a6598e-a874-4094-9338-075331104fd0_at_z6g2000pre.googlegroups.com...
>
> > Sorry if I said somewhere that 'theory was useless,' as soon as I find
> > that post, I'll try to clarify my point.
>
> I'm looking forward to the clarification.

I think it was my other post that was mistaken for my first post in the object oriented thread. I wanted to persist data with transactions without all the 'relational theory' or something to that effect. In the one case that I was asking about, esentially for my purposes relational theory would not be used as a data model. I think someone generalized this to have me mean all relational theory is useless. Probably just poor typing on my part.

The datamodel is essentially ceded to the OO programming language itself, or in terms more familiar to the rdbms folks, there is no datamodel. There is 'new', 'delete', and as these are created, modified and destroyed, they're essentially persisted. Obviously this brings into play everything about what an object is, like shallow or deep copying, all the usual serialization concerns and as I've stated, for a collection I'd like to have something like sql's begin transaction and commit. But otherwise, the persistance layer cares nothing about a datas integrity or consistancy other than the transactions.

And so I don't waste posts, anybody who thinks I'm a troll, I'll probably use this email address, you can killfile me with this. Received on Sun Nov 02 2008 - 16:27:01 CET