[Opensim-users] How to create a corded telephone with a movable cord?
nlin
nlin.message at gmail.com
Wed Mar 28 00:37:53 UTC 2012
2012/3/27, Nebadon Izumi <nebadon2025 at gmail.com>:
> I was going to mention Ninja, but from my testing things can go very wrong
> and bad if the use is not very controlled, like if a ninja linkset goes off
> the edge of the sim, things can go horribly wrong to the point the
> simulator will not start up anymore without editing these lost prims out of
> the database.
Thanks for that report. I hazily recall that there were unresolved
issues with to persisting of in-simulation jointed assemblies, which
is probably related to (if not the same thing as) what needs to be
done when linked assemblies cross sim boundaries. It's an interesting
problem to think about - how would you move one half of an
in-simulation rope into one region while leaving the other half, also
being simulated, in another region? It may be possible with some
fooling around with kinematic constaints on the prims that are at the
region boundary. But then what if the neighbor region goes down, or
even worse, is restored later with a conflicting physical state - one
region says the common segment of the rope is at the NE corner, while
the other region says it's at the SE corner... etc, etc.
> Its certainly something you could experiment with, you just
> need to be very careful, in my sandbox i have put up big walls, and when i
> am playing with Ninja physics i even put up a secondary barrier to capture
> the ninja prims in case things go wonky.
Heh, sounds like a good plan - a containment vessel for exploding prims. :-)
Maybe it would be possible in the code to prevent assemblies from ever
crossing region boundaries unless and until a mechanism is designed to
allow proper crossing. Prims hitting the boundary would bounce back
against an invisible planar constraint. Intuitively, it seems like it
shouldn't be so hard to do, but I haven't looked at the code in a
while...
-Norman (nlin)
More information about the Opensim-users
mailing list