Re: utl_recomp grants

From: Yong Huang <"Yong>
Date: Thu, 26 Apr 2018 15:05:36 +0000 (UTC)
Message-ID: <431387487.1212230.1524755136596_at_mail.yahoo.com>


I agree with Niall. If one object is invalid, that object should be compiled instead of compiling all. I want to add that intentionally invalidating an object sometimes can be used to check for usage prior to dropping the object. Some shops adopt the practice of renaming a table and dropping it some time later, in case the table is still used by an infrequently run application. The same idea can be applied to PL/SQL objects, except you can't rename them. But you can execute dbms_utility.invalidate and later check the object status right before dropping it. If the status changes to valid, you know an obscure program had run the PL/SQL program during this monitoring period. If you compiled it, you would get a false positive in this monitoring.

Yong

> often there's an
> underlying reason why an object is invalid which recompilation won't fix
> and therefore human intervention is required. In addition, the scope of
> your solution recompile *all* invalid objects doesn't match the alert "
> *this* object is invalid" - surely a better approach would be to attempt to
> fix the particular object that is invalid.

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Apr 26 2018 - 17:05:36 CEST

Original text of this message