On Thu, 2003-04-24 at 01:49, Sailesh Krishnamurthy wrote:
> >>>>> "scott" == scott marlowe <scott.marlowe> writes:
>
> scott> Plus the fact that the underlying pg_ tables are stable
> scott> from release to release makes it a bit awkward to upgrade
> scott> the servers they play on. Most of them have gone ahead and
> scott> created views that give them a consistent view of the parts
> scott> of the database they need.
>
> This is exactly the reason why in db2 _no_ guarantees are made
> regarding the constancy of the system catalogs (that are in the SYSIBM
> schema). Instead, the equivalent views (in the SYSCAT schema) are
> _never_ broken (whether in a point release or a new version). In fact,
> the SYSCAT views correspond to the ISO SQL standard.
The INFORMATION_SCHEMA? Out of curiousity, how do they handle DB2
extensions? Do they create new views in that schema? Do they ignore
them?
I'm going with the assumptions DB2 has extended SQL specs in some shape
or form.
--
Rod Taylor <rbt@rbt.ca>
PGP Key: http://www.rbt.ca/rbtpub.asc
