I am starting with a major overhaul to the physics engine today. Since I may have missed some earlier bug reports please post relevant links or screenshots here. Even better: if you have a level that clearly demonstrates the problem send it to mgrubert@conitec.net and I will get on it right away.

after version 6.00.6, physics entity vs. non-physics entity collision wasn't as reliable as before. no matter if my.polygon is on or off for the hit entity, physics entities often rocket around much too extreme after a collision.

it can be seen with the crane simulator for example. try to shoot some balls at the base of the crane with the current engine version!

it would be nice if you could get polygon accurate collision to work with models like the slide i posted in the beta forum some months ago.

...
do you only fix problems or should we also post suggestions for new features?

The only "bug" I can account for 100% atm is the inconsistancy in the PE's handling of the angular variables. Setting torques seems fine, but using phent_getvelocity in a angular context is not fine.

Here's the thread and a more comprehensive discussion of the issue:

http://www.conitecserver.com/ubbthreads/showflat.php?Cat=&Number=288514&page=0&view=collapsed&sb=5&o=31&fpart=1

This bug is at, oh, 60% confidence:

I have my object drop from the sky under the influence of the PE. At one point I attached a Ball Constraint to the player and to the world. Point where it was fixed to the world was a few quants to the right of my object. What I expected to happen, happens...I swing. Problem is, if I take out all drag and friction, I should reach my starting height again, right? Energy conservation and all that muck. Yet each and every time, playing around with all manner of ph_correction and such, I ended up falling "limp", losing my swing, and just hanging out. There is nothing in my code that I can think of that would acount for this. It's almost as if *someone* forgot to WD40 the ball joints!

Finally, this bug is at 40% confidence:

In order to calculate my objects acceleration, I have old and new velocities stored as skills and I then calculate acceleration each timestep. Now, my project is C/S, but I have the server in non-dedicated mode so that it too has a player. What I have noticed, though I'm still not sure it's not MY bug, is that I start falling from the sky with relatively constant acceleration BUT the closer I get to the ground, the more wildly varing values of accel. I get. Since if I"m playing my player on the server, there should be no update lag, and even WITH lag, this selective discrepency is odd. Like I said, not sure that it's endemic to my code or some bug.

• Add support for bones, so that the phyics engine can take controll over certain bones, which are then no longer animated, but simulated.
  
• Add some sort of softbody physics based on springs

EDIT: Ups, I thought that omission means something like suggestion. But since I think the features are imported, I let them be .

• auto rest feature and better stacking performance without jittering
• limits for ball joints
• cylinder and capsule collision hulls
• compound collision objects (fixed joints)
• definable physics properties for individual level geometry surfaces
• fix wheel bending
• fix (sometimes) jumping wheels on flat ground at the edge of two adjacent level polygons

<edit> ...i couldn't resist to post some additional ones! </edit>

does a6 currently use ode 0.035 or 0.039? i read 0.039 is a big improvement but it has been released after the a6 release.

My list of features that I need Implemented.

- Ball Constraint Limits

- Cylinder Hulls

- Fix Wheel Bending(Major)

- Fix sudden erratic behaviour which randomly occurs with physics entities

- While using a wheel constraint physics entities leave its set position, and snaps back. This happens randomly as well, its almost as the wheel sticks to the ground for a fraction of a second then just snaps back. This hinders movement.

- Movement of a vehicle using the physics engine on terrain. Currently when a vehicle moves over a terrain it is tossed all over the place due to the amount of different angles of polygons it is traveling on. This is due to the wheel bending most likely. I also noticed that the wheel sticking problem is also more frequent on terrain.

- Allow events to be used with entities enabled in the physics engine.

The issue I have with the physics engine is the wheel falling through the terrain. I currently made a small fix of this but a real fix would be great. I know that I must send a build of the level with scripts to Conitec, and stated that I would do it before - I was very short on time so I never did re-send it. Next week, I'll send you and jcl the problem so you can review it. I'm sorry about the delay.

BTW, Will my old scripts using the physics engine be compatible once you implement the new version?

Very thankful, and glad to hear you started!

Personally, I would like to see the following- it might restore enough faith to finally upgrade to A6 pro for the physics:

- Make it possible to "clamp" physics entities to bones on a model- it would save me weeks of coding and longer in debugging for my ragdoll physics (which check out on the concept models finally!!).

- Fix the axle bending problem

- Physics event handling (Major!)

-Rhuarc

Hehe, if you change anything on ODE itself be sure to post it on ODEs pages

~Phantom88~

in 6.22 physics collision problems seem to be much more extreme at level polygon edges (but maybe i just didn't test that in earlier versions). marco please try to shot balls at the red debug mode lines in the crane level. the balls shouldn't reflect in these directions on a flat ground!

does the physics engine use the tesselated bsp geometry for collision? wouldn't it be better (less edges) if just the polygons from the blocks in the .wmp file were used?

...
could you reproduce the collision problems with shooting balls at the crane base now? this problem makes the physics engine unusable because it doesn't only happen with the crane but with most models on my pc!

After alot of work, I have effectively singled out exactly why certain physics entities have a tendancy to have bad collision on terrain! A while ago I posted about bad collision with physics entities and terrain. http://www.conitecserver.com/ubbthreads/showflat.php?Cat=&Number=320317&page=5&view=collapsed&sb=5&o=&fpart=1#320317 You also may remember that I had previously believed the problem had to do with my function names - I was wrong. After much testing, I've come to the the conclusion that the last entity enabled in the physics engine will have faulty collision detection on terrain! Whats weird is that the entites which previously had faulty collision, will suddenly have perfect collision detection, when another entity is enabled into 3dgsPE on terrain. This problem is much more wide spread then I first believed. Upon further testing I noticed that no matter what, whether there is an constraint or no constraint the last entity enabled into 3dgs PE will have faulty collision detection with terrain. Hopefully this will be an easy fix for you Marco. Simply check why the engine gives faulty collision detection between terrain and models which were enabled in the physics engine last. If you have any questions please ask - I know a few others had this problem, and I'll soon release the vehicle code with the "cheap fix".