A collision bug ?

[forestluo]

I created two box entities. One was static (mass = 0.0), the other was movable (mass= 10.0). The static one is lower and the movable one is higher (2.0 unit higher along y axis). They were both 1unit * 1unit squre box (along x axis and z axis). But they are all thin, only have 0.01unit thick (along y axis).

I found the following results :

1) If they were put parrelly (vertical to y axis) and collided with each other with gravity force (along y axis). They would directly cross each other. Just as nothing was happened.
I had thought that they would have collided and the movable one would have jumpped back.
2) If I changed some angles of one entitiy, the simulation woule become correct . The statcic one would resist the movable one, and the collision would happen its uppder surface.

I also created a thin square static slice array. All are 1 unit * 1 unit, leave each other with 0.01 unit spacing. They all only have 0.01 unit thick. Once I collided them with a single thin squre slice, some errors (according to a personal experence in the real world) would be seen directly.

I reported those just according to my exprience. I don't know whether other person has already done the same expriments. Maybe this is a known problem.

[JernejL]

With very thin slice objects, minor mathematical errors might show up which are a result of computer floating point imperfection.

As for the head-on direct collision bug, it's possible two perfectly aligned rectangles don't collide properly, but only julio can look into that.