Re: [gl4java-usergroup] OpenGL vs Java3D (was: Re: GL4Java FAQ]

[Kerry L. B. <ke...@vs...>]

At 02:05 PM 4/13/00 -0400, RICK ANDERSON wrote:
>
>It also gets back to my original point, I assumed that the folks who did
GL4Java
>had a reason for the decision/approach that was made.  In my case, I guessed
>wrong on the reason.
>
>Either way, I'd suggest capturing, IN THE FAQ, the essence of the reason for
>putting a thin layer on top of OpenGL as opposed to using the Java3D API.
There
>are d*mn good reasons for the choice and it makes sense to capture them.

FWIW: It wasn't my decision, but when I found gl4java I was just about to
do the same thing.  My reason was simple - there are many conflicting
design requirements in 3D apps - from games to widget spinners.  The only
way to satisfy both ends of the API spectrum is to provide both ends of the
API - low level and high level, preferably in a manner which allows
coexistence.  OpenGL is the best low level cross platform rendering
abstraction, so it made sense to start there.  Thanks to Sven (et. al), I'm
getting to focus on my high level and application code.

>Are you the guys who presented "How to keep 10,000 online gamers happy" last
>year at JavaOne?  There was an awful lot of "then a miracle happens and we
won't
>tell you how" in that presentation :-)

I didn't present there, done plenty of other presentations in the past.
That group is typical - they think dynamic load balancing is sufficient.
It isn't, its caps out at a few thousand, breaks well before 10k.  Properly
implemented "adaptive clustering" has no theoretical bound, it approaches
O(n).  I'll file preemptive patent soon on the stuff I'll be releasing.
Not to mention all the security issues that have to be done "RIGHT".  Wish
I could say more, am willing to answer most specific questions off list. :)




Kerry L. Bonin
Sr. Engineer, Security/Cryptography & Advanced Visualization, Cisco Systems.
VScape lead architect - Adaptive secure clustering for multiuser VR.