manifold points and constraints

[GoogleCodeExporter]

What steps will reproduce the problem?
I think there is a bug in applying forces stored in contact points when I 
connect two bodies with a joint, see the attached reproduction case.
There is one box on the top of the other. When you press the 'e' key, they get 
connected by a constraint, see the activateJoint method.
When you press 'e' again, they get disconnected and a torque is applied on the 
top body for one physics step.
When you press 'e' three times quickly at the beginning, the top body maintains 
rotating.
1. The first 'e' connects the bodies. 
2. The second 'e' disconnects them, makes the top body to rotate and fills the 
contact points with some force values. 
3. The third 'e' connects the bodies again, but the top body keeps rotating, 
although it should stop.
I think the forces in contact points are still applied.

What is the expected output? What do you see instead?
The top body should stop rotating. It keeps rotating.

What version of the product are you using? On what operating system?
The latest revision 2639.

Please provide any additional information below.
I provide an artificial example just to reproduce it, but the same happens to 
me with a point-to-point constraint.

Original issue reported on code.google.com by bol...@centrum.cz on 15 Apr 2013 at 10:15

Attachments:

• BasicDemo.zip

[GoogleCodeExporter]

The contacts points are generated, but the hinge constraints seems to cancel 
the gravity, so the upper box 'floats' above the lower box, hence no friction.

I don't think it is a bug. If you disagree, please re-open a new issue and 
explain further (need to clear the bug tracker)

Original comment by erwin.coumans on 10 Sep 2013 at 8:51

• Changed state: WontFix