Re: Mixing OO and DB

From: JOG <jog_at_cs.nott.ac.uk>
Date: Sun, 23 Mar 2008 18:39:49 -0700 (PDT)
Message-ID: <78b8d233-984e-4ce1-8f44-7d3d4b65fe4d@n58g2000hsf.googlegroups.com>


On Mar 22, 2:28 pm, "Dmitry A. Kazakov" <mail..._at_dmitry-kazakov.de> wrote:
> On Fri, 21 Mar 2008 22:42:44 -0700 (PDT), frebe wrote:
> >> The claim that the application can't do translations is
> >> incorrect, on its face. SQL isn't going to return DAGs, or even
> >> queues, stacks, or maps.
>
> > A DAG can obviously be represented as a set of tuples.
>
> LOL. For that matter, a DAG can be represented by a natural number. This by
> no means imply that ALU were operating DAGs, or that the ADD instruction
> would return DAGs. You could certainly represent DAG by used paper cups...

You seemed to have missed the point that a DAG _is_ a set of tuples. To a mathematician, creating a DAG using pointers and memory addresses isn''t much different from using paper cups connected by bits of string.

>
> > The same
> > applies to queues, stacks or maps, even though I can't really see why
> > you need a low-level data structure like a map, when relations are
> > availible.
>
> Because an algebraic structure is not just a set. Isomorphism of elements
> does not make algebraic structures same. Is it so difficult to grasp?
>
> P.S. To the list of problems I put before you to solve:
>
> (Message ID: 16zwfmkrguxvp.xiq5rgz36sbc...._at_40tude.net)
>
> 10. Let G be a DAG. Compute G* (transitive closure).

Just use a recursive join. I fail to see the big deal.

>
> --
> Regards,
> Dmitry A. Kazakovhttp://www.dmitry-kazakov.de
Received on Sun Mar 23 2008 - 20:39:49 CDT

Original text of this message