Mladen Gogala wrote:
> On Sun, 23 Apr 2006 02:20:10 +0200, Maxim Demenko wrote:
>
>> Mladen Gogala schrieb:
>>> In Oracle RDBMS 10.2 it is practically impossible to invalidate a PL/SQL
>>> object. I tried with adding indexes, dropping primary key
>> Not every ddl invalidate the dependent procedures, only changing layout
>> of tables ( i.e. add/drop/modify columns, maybe something else, but not
>> add/drop constraint, alter table move etc.)
>
> Those DDL can change plans of the underlying SQL statements and,
> therefore, should invalidate the dependent procedures.
>
>
>> and analyzing
>>> the underlying tables,
>> Here you are joking - right ?
>
> Nope. No jokes here. Analyzing the underlying tables can also drastically
> change plans and should, therefore, invalidate the procedures.
>
>
>> but the PL/SQL procedure build on top of the table
>>> remained valid. Oracle 9i PL/SQL objects were much more sensitive.
>> You probably don't have 9i instance right now to test, the behaviour is
>> the same in 9i and even 8i.
>
> You are right, I don't have access right now. I will check it on Monday,
> though.
>
>
I fail to see why changing execution plans should
invalidate functions and procedures?!?
What does the state of a procedure have to do with execution plans?
Or, to take it even further, SQL with PL/SQL?
--
Regards,
Frank van Bortel
Top-posting is one way to shut me up...
Received on Sun Apr 23 2006 - 06:16:16 CDT
