jME roots ... MMOGs?

Hello,



I've just begun to look at jME … noticed that jME has some of its roots in David Eberly's books and libraries.

brcain said:

Does the JNI boundary eventually become an issue?


No. Unless you want the utmost performance, it's not really an issue. There's a slight overhead. But especially for modern videocards, if you're doing a lot of jni (and thus OpenGL) calls, you're probably doing something wrong anyway. Even then we're still usually talking single digit percentages in performance differences.

There's some other areas where Java performance might lag behind a bit behind C/C+ (espc raw byte manipulation, because the bytes you want to manipulate are often in native buffers, which you can only access through wrapped methods in Java), but not by that much. In the worst case you can always do that stuff in C/C++ and add.. a jni call.
llama said:

No. Unless you want the utmost performance, it's not really an issue. There's a slight overhead.


Glad to hear.  I realize much of the issue is not due to the culled render list ... but rather in terms of network software architecture, seamless database paging, etc.

I'm curious to see what jME provides and what is left to the developer.

I'm assuming from your response there's no issue with heavy usage of shaders either.

Thanks,
  Ben