Re: Update through views?

On Wed, 7 Mar 2001, Tom Lane wrote:

> I can guarantee that we won't ever support these things on an arbitrary
> view; there'll always be a need for hand-made rules in cases where the
> system is too dumb to figure out a plausible rule for updating a view,
> but the programmer knows what he wants to have happen.

As Richard Huxton pointed out.  I was taking a very simplistic view of
views. :-)

>
> The SQL spec calls out a set of conditions for a view being "updatable",
> which essentially means that the view is simple enough that a DBMS
> should be able to derive update rules for it automatically.  At some
> point we'll probably try to build machinery to implement automatic rule
> creation for those sorts of views.

Sounds good but given my now enlightened view I'd probably not use it.

> But it'll always be possible to write your own rules if you don't like
> the automatic ones or the view is too complex for automatic rule
> creation.

As someone stated (I think Larry Wall) if programming were easy it would
take something as complicated as a human to do it.


Thanks for the information.


Rod
--