Re: 10053 trace for sql fired from pl/sql (stored code)

Hi,

> v$sql_plan_statistics already IS another sql). I believe sqlplus keeps the
> cursor of the most recent sql statement open, so it may be that the cursor
> needs to be closed for the statistics to be "populated". But then these
> are v$ views, so there is nothing to BE populated.

For scalability concerns some execution statistics are returned and added up to v$sql stuff only when cursor is closed, otherwise we'd have to go back to the shared cursor every small operation (like fetch etc).

Some of the stuff in X$KGLOB is incremented latchless way as well I believe.

Tanel.

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