Re[2]: [ODE] CVS + OpCode buggy?

[Nguyen Binh]

MB> Yep, I've compiled OPCODE. I have since downloaded the latest stable 0.039 to 
MB> try that out, but the problem persists. test_crash always segfaults when I try 
MB> to switch to stepfast both with and without OPCODE compiled in. however, the 
MB> stable version no longer seems to have problems with test_ode, and it doesn't 
MB> force you to use single precision with OPCODE the way the cvs does. As far as 
MB> test_crash goes, I'm really not sure what the deal is. Consistently, it starts 
MB> up, i press 'f', and boom - segfault. Getting rid of OPCODE doesn't seem to 
MB> make any difference. However, I haven't yet tried writing any ode code of my 
MB> own to see if the stepfast problem persists. Kind of discouraging right off the 
MB> bat, though :(

    That's segfault is not stepfast's fault.On test_crash demo,
    there are so many bodies (nearly 1000 ?), that's OK for stepfast
    which require only O(n) amount of memory (n = number of body), but
    not really OK for traditional LCP solver which require much more
    stack memory to run. So the problem here is stack overflow.

    Solution :
             + Reduce number of boxes.
             + Increase prog's stack size by some compiler settings....
             
--
Best regards,

---------------------------------------------------------------------
   Nguyen Binh
   Software Engineer
   Glass Egg Digital Media
   
   E.Town Building  
   7th Floor, 364 CongHoa Street
   Tan Binh District,
   HoChiMinh City,
   VietNam,

   Phone : +84 8 8109018
   Fax   : +84 8 8109013

     www.glassegg.com
---------------------------------------------------------------------