[ODE] Precision specific functions
starkos at gmail.com
Fri Sep 1 08:45:38 MST 2006
On 9/1/06, Jason Perkins <starkos at gmail.com> wrote:
> A while back, there was a brief discussion about introducing
> OpenGL-style precision specific methods to ODE. That is, *in addition
> to* the existing dBodyGetPosition() there would be:
> void dBodyGetPositionf(dBodyID body, float* pos);
> void dBodyGetPositiond(dBodyID body, double* pos);
> The purpose of this is to ease integration with third-party code. Is
> there a general interest in this approach, or will it just clutter up
> the API?
Maybe I should explain why I am asking about this. I'm bothered by the
fact that the numeric precision chosen for the solver bubbles up into
the client application. This doesn't feel appropriate or necessary; my
application shouldn't know or care what ODE is doing internally.
In my specific case, I need a .NET wrapper for ODE that doesn't
require managed C++ or unsafe code. In this particular application,
positions are stored in doubles and everything else is float.
More information about the ODE