Something rarely causes world to be null during world generation. Extabiomes cannot assume that the pointer is non null.

We can write a message to the log if world is null..and then when vanilla code bombs out it will dump a stack trace. If it doesn't bomb, then it's really a non event.

I'd recommend logging any such occurrence, so people can locate the cause.

Does java have a way to say "Dump the current stack trace for debugging"?