On Tue, Jul 17, 2001 at 07:13:10PM -0400, Tom Lane wrote:
>
> Nonetheless, a properly indexed pg_depend table would allow you to find
> these objects directly, and again to find their dependents directly,
> etc. The brute force approach would require a rather expensive scan
> over all the system catalogs, plus nontrivial analysis for some types
> of system objects such as functions. Repeating that for each cascaded
> delete is even less appetizing than doing it once.
Stated that way, the performance argument sounds very convincing. However,
the _real_ convincer for me is the support for user designated
dependencies, as Tom pointed out earlier. That allows the system to do
as much as possible automatically, (even functional dependency analysis,
if someone want to write it) but doesn't require the automatic mechanisms
to be perfect: the DBA has a mechanism to do the crazy, edge case things.
Ross
