>if I could get into the TRY section of the PG_CATCH()/PG_TRY()> construct without an intervening elog(ERROR) then I'd have a> chance ...
Sorry, I meant "the CATCH section of the PG_TRY()/PG_CATCH()construct."