"Leyne, Sean" <Sean@...> wrote:

>
> As such, the ability to mark on object as invalid would be "a good
> thing".

While I agree about invalidation of metadata objects, cached statements are
a bit different beast. Nothing depends on them, actually. For me, it doesn't
really matter whether they're invalidated (i.e. defered to be recompiled
upon next request) or just dropped from cache immediately. This looks like
an implementation detail at the first glance. Metadata changes are the
reason for this to happen. I'd also expect significant data distribution
(statistics) change and/or index DDL to be such reasons.


Dmitry