Dan,

known bug, Oracle is not implementing workaround and IBM won't fix the
bug. This almost always happens on e= field. I haven't seen too many
instances of this on 10g and aix 5.3/5.4 but will keep an eye out.

That bug afaik is not going to get fixed anytime soon, so I wouldn't
keep high hopes. We even checked with IBM. Couple of years ago when we
first encountered this (posting in oracle-l dates to 2004), I used to
have a complete list of responses from Oracle and IBM. Probably saved
somewhere on my HDD but definitely not handy.

ps: I have seen this error from 9012/aix 5.1 onwards. If you want I'll
look up the details, but no guarantees
Raj
--
http://www.freelists.org/webpage/oracle-l