Re: Interested in a moderated theory forum?

"Costin Cozianu" <c_cozianu_at_hotmail.com> wrote in message news:buhi7n$hm1gm$1_at_ID-152540.news.uni-berlin.de...
> Marshall Spight wrote:
> > "Costin Cozianu" <c_cozianu_at_hotmail.com> wrote in message
news:buhbtp$bt01j$1_at_ID-152540.news.uni-berlin.de...
> >
> >>If you disagree, please show me a post on comp.database.theory that can
> >>have some reasonable claim that scientific knowledge was being
> >>discovered and validated in this forum and state of the art advanced,
> >>cause otherwise I can show you plenty of amateurish psots :)
> >
> >
> > Attempts to advance the state of the art here are generally shot down
fast.
> >
>
> No kidding. But if you shoot them down fast, then that's a great service
> you do.
>
> Remeber http://www.c2.com/cgi/wiki?CriticsAreYourBestFriends

Yep.

> > That aside, I'll nominate the discussion of atomic updates as an
> > alternative to current-style transactions as an example of an
> > attempt to advance the state of the art through the application
> > of science, or at least math.
> >
> I would nominate that discussion as a good example of half baked ideas,
> for which arguments are made in ignorance and in disconnect with the
> current body of scientific knowledge (transaction theory, google for
> Weikum and Vossen "Transactional Information Systems"), which,
> regardless of the current solutions, has largely and correctly
> identified all the problems that transaction theory needs to address.
>
> Those problems that a new theory of transactions (or theory of databases
> without transactions), have not even been systematically identified and
> specified, much less address by that thread.

Well from my side of the fence, I got some reasonable criticism, none of which shot down (for me at the least) the core of my idea - that being that transactions are harmful to the (logical) database model.

None of my friendly critics in this group managed to identify systematically any 'problems' with a theory of databases without transactions. OK, so I was trying to lay the burden of proof on the other side - I see no problem in constructing a useful theory without transactions, so those in favour tell me why they are necessary. No one could IIRC.

Am I in ignorance and in disconnect? Well we all are to some degree or other. Have I studied Weikum and Vossen "Transactional Information Systems"? Well no. Do I even own a copy of Gray? Well no. Have I flicked through such books? Yes. Do I still consider transactions to be a 'bill of goods' yes? Am I the only one of that opinion? No. If I had more time would I study more carefully, Yes. Would I like to publish peer-reviewed papers? Hummm. Would I like to test my thoery with some real software. You bet ya. Is there much immediate hope of that? Nope.

> Had the proposers tried to
> publish their work in an academic settings '''and''' test their theory
> with some real software, I have no doubts a better output would have
> ensued, even if that output might have been a dismissal of their ideas.

Yes but that does take a smidgen longer than firing off a few newsgroup messages.

> It's much like the relation between "The Third Manifesto" type systems
> and the current body of knowledge identified as Type Theory.

Humm, I think it's about time we had a thread focused on that one.

> Best,
> Costin

Regards
Paul Vernon
Business Intelligence, IBM Global Services Received on Tue Jan 20 2004 - 21:08:38 CET