Re: [HACKERS] Re: [SQL] Proposed Changes to PostgreSQL

Don Baccus wrote:
>
> >
> >A good thought, but we still need one good front end that supports
> >all the features.
>
> I wasn't think in terms of this being mutually exclusive with your
> desires. Merely raising up the notion that the possibility exists
> of creating a sandbox, so to speak, for people to play in, a tool
> for the exploration of such concepts.

So we would be returning to roots. The original Postgres was exactly that -
a tool for the exploration of such concepts.

> No, and I wasn't arguing that you shouldn't move forward, either. I
> was just stating my personal opinion regarding the utility of simple
> OO-ish features, that's all.

Yes, it needs quite much discussion/design befor going forth, lest we
will be in the next level of the current situation where some peoples
usage of the current limited inheritance is an obstacle to moving
forward to a more developed one.

> >> One reason I raise the issue of possible multiple front-ends (or making
> >> it easy for folks to make there own by making the parser->optimizer/backend
> >> interface more general) is that this whole area would seem to be one
> >> that begs for RESEARCH and experimentalism.
> >
> >No research is required. I simply want to implement the ODMG STANDARD
> >for ODBMS databases on PostgreSQL. There are no great design issues
> >here, just a matter of nailing down the details so that everyone can
> >live with them.
>
> Well...that's sorta like saying no research into procedural language
> design is needed 'cause now we've got C++.
>
> Whether or not the existing standard for ODBMS is the greatest thing
> since sliced bread, I find it hard to believe that no research is
> required or design issues raised by the fundamental problems of
> database technology.
>
> Maybe I'm wrong, though, maybe the problem's been solved.
>

My wife has forbidden me to buy any sliced bread, because the slices are of
wrong thickness.

Hardly the situation can be any better in OODB design.

The ODMG standard may be a good starting point for discussion, but one can't
run any programs on a standard - one needs a real db.
And IIRC the standard is only semi-public (not freely
available/distributable).

------------------
Hannu