RE: Unshared cursors redux

From: Boris Dali <boris_dali_at_yahoo.ca>
Date: Fri, 23 Feb 2007 12:20:32 -0500 (EST)
Message-ID: <385960.41988.qm@web32406.mail.mud.yahoo.com>

Rich,

We discussed this issue on this list back in 2003. I think the answer is no, child cursors created due to different bind lengths are **not** flagged in v$sql_shared_cursor in contrast to the case of using different datatypes that does lead to raising this bind_mismatch flag.

Here's the top-level link for that discussion (and from there you can folow the whole thread):

http://www.mail-archive.com/oracle-l@fatcity.com/msg88023.html

I didn't follow this thread closely, but if you are on 10.2.0.3 you probably know that there's an infomous one-off patch# 5705795 (for Linux) that you might want to look into

Thanks,
Boris Dali.

Rich Jesse <rjoralist_at_society.servebeer.com> wrote:

> So, based on that, I would
> expect that the absence of explicit binds, along
> with NOT using the dreaded
> CURSOR_SHARING=FORCE|SIMILAR init.ora parameter, in
> a cursor would cause
> that cursor to either be shared or to have a reason
> in V$SQL_SHARED_CURSOR
> as to why it would not be shared. But binds for
> different executions that
> cross allocation sizes would seem to be the
> definition for the
> "BIND_MISMATCH" column of V$SQL_SHARED_CURSOR,
> wouldn't it?

....

Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

--
http://www.freelists.org/webpage/oracle-l

Received on Fri Feb 23 2007 - 11:20:32 CST