Re: The Practical Benefits of the Relational Model

"Leandro Guimarães Faria Corsetti Dutra" <lgcdutra_at_terra.com.br> wrote in message news:aohbif$mkd5c$1_at_ID-148886.news.dfncis.de...
> Nathan Allan wrote:
> >
> > OO Prescriptions 4 and 5:
> > Transaction support in the current version is limited to the
> > supporting storage engine.
> > In the next version we will implement full transaction support as
> > prescribed by the Manifesto.
>
> Including Date's new "atomic statements only" idea, meaning that each
> statement should be its own transaction, by doing several actions in
> parallel?
>
> I do think this, together with the default values proposition, to be
> the most radical part of the TTM. Not that I reject any.

What would be more 'radical' would be to recognise that not only are transactions not logically required, they are positively harmful to any RDBMS that purports to support databases that are truly application independent.

Transactions are absolutely not things that *users* should be exposed to. IMO, it is one of the (admittedly) few errors that Date and Darwen make in the manifesto, but, again IMO, possibly the one that has the potentual to cause the most problems if implemented.

Regards
Paul Vernon
Business Intelligence, IBM Global Services Received on Tue Oct 15 2002 - 18:34:31 CEST