Yong, thank you for your reply. Your parse_failure.txt shows a similar question. My initial question was to distinguish between the real ORA-942 and something like "object exists, but user does not have sufficient grants". I'd say this IS possible. - So an educated researcher can get more informations about an database than the pure grant-setup would allow. My 2nd question is to explain when "recursive calls" and "execute count" are increased. That's still not consistent for me. Maybe I should address it in a 2nd thread. Martin On Sat, Sep 15, 2012 at 5:07 PM, Yong Huang <yong321@xxxxxxxxx> wrote: >> Is it somehow reasonable to tell the existence of an object by the higher >> "recursive calls" for a ORA-00942 on an existing object (in comparison >> to a non-existing object) ? > > Martin, > > So you want to infer there's ORA-942 (table or view does not exist) from > statistics. You may also want to take in account the session's "parse > count (failures)" statistic, which of course increments in other cases than > ORA-942. My note > lists different scenarios (see section "What errors cause parse failures?")" > http://yong321.freeshell.org/oranotes/parse_failure.txt > > And, when you test your session's "recursive calls" and "execute counts", you > may want to monitor this session from another session instead of querying > v$mystat which itself causes these stats to increment. > > To accurately record ORA-942 on the server side (not more not less), you can > just set event 942 errorstack at a low level. > > Yong Huang -- //www.freelists.org/webpage/oracle-l