Re: Describing the Janus

David Cressey wrote:
> I want to describe an object that I've been imagining for a while. I've
> called it the "Janus".
>
> A Janus has two faces. I can describe each face reasonably simply: one
> face is an SQL client, the other face is an object, interacting with other
> objects in an object world.
<snip>
>
> Is there any use for such a thing? I imagine that it might be an approach
> to solving the impedance mismatch between the (SQL based) relational DBMS
> products and the world of Object Oriented application development.

have a look at Ruby On Rails. you'll need to get used to Ruby to actually use it and all. but it is based on an idea, (so they say) from Martin Fowler (about whom i have mixed feelings), called ActiveRecord in   RoR. from what i've read, and this is only over the last couple of days, ActiveRecord gainsays that there is an impedence mismatch (java folk just luv to make this complaint). and this is what i've always felt.

the rationale: no matter the number of instances, a real computer only stores one text segment (Intel term); so, each method of each loaded class exists but once. the data for each class is stored separately. given this, then the RDBMS fills the instance (and class) data requirement. each composite or inherited class works the same way.

this is not unprecedented. in the Prolog world, which is based on the notion of rule databases (memory resident structures), there is a product called Amzi! (that's not a typo) which maps the Prolog rule database to table rows.

if someone can show that OO databases store method text for each instance, i'd be interested to know. and why. certainly seems pointless.

Bob Received on Mon Sep 12 2005 - 04:34:03 CEST