Re: same query, different parsing users, different plans ...

From: joel garry <joel-garry_at_home.com>
Date: Fri, 6 Jun 2008 12:58:04 -0700 (PDT)
Message-ID: <b18b3295-9d25-4530-abe9-002675a44680@d19g2000prm.googlegroups.com>


On Jun 6, 4:08 am, "stephen O'D" <stephen.odonn..._at_gmail.com> wrote:
> > I will update again if this fixes it.
>
> Giving the 'merge any view' privilege to the non-object owners fixed
> this problem.
>
> Personally I think this is a bug waiting to happen, I mean who would
> possibly have guessed of this sort of behaviour?  The merge any view
> privilege is a feature to enhance security, but the fact that plans
> generated by user_1 are sub-optimal because of it, and then SHARED is
> totally unexpected behaviour.

Thanks for the update (and you too, Charles).

Strange: searching the docs for merge any view says it finds one match in 10 and 11 docs, then clicking on one it says no exact matches were found and gives a bunch. Searching for "merge any view" also says it finds one match, but actually shows one when you click on it. I do not understand the search function on tahiti.

Not behavior to be guessed, just a lesson to be learned, as the not-a- bug shows. I guess.

It's better than this sort of behavior (the old wrong-schema bug): http://groups.google.com/group/comp.databases.oracle.server/browse_thread/thread/3ee2ecec9692457d

jg

--
@home.com is bogus.
“It's amazing how friendly people are when they're naked.” -
http://www.signonsandiego.com/news/northcounty/20080606-9999-1n6nude.html
Received on Fri Jun 06 2008 - 14:58:04 CDT

Original text of this message