From: Ville V. <vil...@gm...> - 2009-04-02 08:50:30
|
>> This correlates with my findings, with a 50-50 split between >> exception/no-exception cases. Thus it seems Note that the virtual call findings do _not_ correlate. When I made sure that the compiler cannot know the real type, virtual calls were *much* slower than even catching ClassCastException. Instanceof should be faster still. That's why I said the following: > With the 50-50 split, it also looks like java VM somehow optimizes > ClassCastException handling. I haven't |