#include   <iostream>  
  class   Controller  
  {  
  public:  
        void   print()  
        {   
             std::cout<<"good";  
        }  
  };  
  int   main(int   argc,   _TCHAR*   argv[])  
  {  
        Controller   *pCtrl;  
        pCtrl->print();  
        getchar();  
        return   0;  
  }

====

^_^, it runs good.

晚上肚子饿了，正好有公司发的月饼，吃了一个… 又要中秋节了。不知道是我最近厌食，还是月饼包装太漂亮，不太入口味…

吃着吃着就有点想家了，貌似好久没和父母一起过这些杂七杂八的节日了，一个人在北京漂泊这么久了，有时候真想tmd一甩头，跑回浙江算了！哎…

年轻人总号称有追求，也不知道我追求到什么了。。。

日子一天一天过，搞不定什么时候一回头，发现自己已经年纪一大把！

1.add multi-language support
2.jsr239(OpenGL ES) alpha

weapon system is done, depending on self-calculated skinned bones. the weapon’s movement is referring to the wrist bone.

XXXXXXXXXXXXX图片删了XXXXXXXXXXXXXXXXXXXX

with aura effects,

XXXXXXXXXXXXX图片删了XXXXXXXXXXXXXXXXXXXX

Cost too much memory!

the player actor:

[frames]  [cache interval]   [vtx count]  [xyz]  [short]
 10000   /        32           *    340       *  3    *  2 = 637500  = 622.5k

one vertex cost:

10000   /        32           *      1         *  3    *  2 = 1875     = 1.83k

less vertices & polygons….. less frames

If you mix 2d & 3d rendering on Hardware Accelerate phone, it will cut down the fps a lot.
On my test on Nokia N82, it costs nearly 10 fps after drawing a 2d fullscreen image.
So I use a 3d plane near the camera instead.

XXXXXXXXXXXXX图片删了XXXXXXXXXXXXXXXXXXXX

self-calculated skinned mesh is done.

from now on, all rendering in the engine is only to call one m3g api:

    g3d.render( vtxBuffer, idxBuffer[0], appearance[0], transform );

the first benefit is get a more fastern shadow,  but there are more float operations… I’ll test it on Monday in details.
Maybe caching all the key-frames is the last optimization way.

XXXXXXXXXXXXX图片删了XXXXXXXXXXXXXXXXXXXX

The rendering will get an incredible speed up! That’s awesome!

(ah, if we manage all the models in our own format, and rebuild them in one at last before flush to the api? — maybe we can have a try…)

XXXXXXXXXXXXX图片删了XXXXXXXXXXXXXXXXXXXX