What steps will reproduce the problem?
Replace ConstraintDemo.cpp with version in linked thread and fire cubes at the
centre of the constraint array.
What is the expected output? What do you see instead?
The angular motors should kill the velocity in the 6dof constraints, instead
the entire system becomes unstable.
What version of the product are you using? On what operating system?
Trunk on Linux 64bit.
Please provide any additional information below.
The problem stems from the fact that btSequentialImpulseConstraintSolver does
not provide option to randomize non-contact constraints, if support for this is
added (patch attached) and randomize constraints is enabled the system behaves
as expected.
Note that the ODE quickstep solver provides this as an option and was found to
exhibit the same problems when used with bullet and constraint randomization
disabled.
See thread here:
http://bulletphysics.org/Bullet/phpBB3/viewtopic.php?t=7672
Original issue reported on code.google.com by peter.kyme on 8 Dec 2011 at 5:14
Original issue reported on code.google.com by
peter.kyme
on 8 Dec 2011 at 5:14Attachments: