I'm still wondering if actual object pointer will ever get deleted, but my fix did not break that if it did not fix that either (will broadcasting actually get returned to RTIG so object will be deleted correctly). ObjectSet's killFederate should have delete if it even does anything in kill now that this fix will handle all objects in ObjectClassSet's killFederate.
Anyways I think this fix should be in next release whenever it will be released.
I decided to test CERTI again after noticing that Portico had bad message order issues when transferring object attribute ownerships (and also ridiculous 1 sec slowdowns with unowned attributes). CERTI seems to handle ownership transfers flawlessly :)
Also CERTI source code seems to be quite readable as I managed to fix this bug in couple of hours without ever reading CERTI's source before.
- Timi Tuohenmaa