Sorry, the question is: Is anyone experiencing similar changes in behavior from 3.1 to 3.2?
ORIGINAL POST: As I posted in the Unity Community... Unity update 3.2/3.3 has rendered my project practically useless.
The short of it: One improvement I see on 3.2/3.3 is that the behavior of objects that used to blow off the screen never to be found again has ceased.
On 3.1 it is a rare occurrence when the vehicle hangs on the terrain, wedged against a vertical... usually only if jagged terrain structure is present.
On 3.2/3.3 it is a constant nuisance.
The detail if you care to read: Gameobjects climb up cliffs or terrain walls on their own terms. The objects can be of any size as far as I have tested. I can play in the scene for an extend period of time and get back to the location of objects that are still hung/velcroed into the side of verticals orders of time later.
Same project in 3.1 does not exhibit such behavior… objects quickly drop to the low terrain as expected.
3.2/3.3 this 'magnetic' effect takes a 1200kg vehicle, dances it up and down a wall, holds it sideways(as if velcroed) yet dangling by a thread, with engine killed(no code executes to drive wheels).
Sometimes I can drive the vehicle on the side of the cliff/wall as if on flat land, sometimes it comes down back to the flat part of the terrain, more than often it hangs indefinitely at a 90% angle(on its side) about a meter+ up on cliff requiring a restart or a game local reset.(note: it is not always 90%)
Sometimes I am able to drive it back down(sideways, lying on its side).
After a COLD reboot it seemed I had improved behavior, but not so... problem persists. Objects that get near a vertical can shoot up, then fall back, and stop before reaching flat terrain, and come back up on their own term and slide back down again, sometimes hang indefinitely before reaching flat land or sometimes reach flat land and roll. I think I am observing that the velcro effect occurs where the vertical meets a part of the vertical that is embossed further out but not a ledge where you would expect something to hang or come to rest.
Whether it is a 3.2/3.3 specific problem or a terrain conversion problem between 3.1 and 3.2/3.3, I am not sure.
[Same vehicle, same physics]
I do know that a new scene created in 3.2/3.3 with a minimal terrain(an 8m wall raised up from the base flat terrain) the velcroed behavior persists perhaps to a lesser extent.
I am unable to determine if it is because the terrain is different in 3.2/3.3 or a problem in the conversion of the terrain, or if the design for each test is not exactly the same.
I suspect it is a combination.
On the 3.2/3.3 I can drive the vehicle straight up the 8m vertical(less than 1m wide) and roll on thru to the other side, or stop at the top and drive on the narrow plateau...
On 3.1, the vehicle is not able to climb that far up only far enough to cause the scene to start shaking and the vehicle to exhibit behavior of heavy struggle while looping back down.
Short of redesigning the entire project in Unity 3.3, from naked terrain, I see no way to test any further. That would take an enormous amount of time, especially recreating the terrain, and it would be impossible to redesign the terrain exactly the same, thus many areas in the game would have to be adjusted to the new terrain... and if the behavior persists, it would be a MASSIVE amount of time loss. <- only one word for that: NIGHTMARE!
I have added weight to the vehicle, to no avail.
I hope someone else is experiencing these effects and a fix is applied.